MNRAS 500, 2359–2379 (2021) doi:10.1093/mnras/staa2246 Advance Access publication 2020 August 18

Bar effect on gas-phase abundance gradients. I. Data sample and chemical abundances

A. Zurita ,1,2‹ E. Florido,1,2‹ F. Bresolin ,3‹ E. Perez-Montero´ 4 and I. Perez´ 1,2 1Departamento de F´ısica Teorica´ y del Cosmos, Campus de Fuentenueva, Edificio Mecenas, Universidad de Granada, E-18071 Granada, Spain 2 Instituto Carlos I de F´ısica Teorica´ y Computacional, Facultad de Ciencias, E-18071 Granada, Spain Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 3Institute for Astronomy, 2680 Woodlawn Drive, Honolulu, HI 96822, USA 4Instituto de Astrof´ısica de Andaluc´ıa, Camino Bajo de Huetor´ s/n, Aptdo. 3004, E-18080 Granada, Spain

Accepted 2020 July 17. Received 2020 July 14; in original form 2020 April 29

ABSTRACT Studies of gas-phase radial metallicity profiles in spirals published in the last decade have diminished the importance of galactic bars as agents that mix and flatten the profiles, contradicting results obtained in the 1990s. We have collected a large sample of 2831 published H II region emission-line fluxes in 51 nearby galaxies, including objects both with and without the presence of a bar, with the aim of revisiting the issue of whether bars affect the radial metal distribution in spirals. In this first paper of a series of two, we present the galaxy and the H II region samples. The methodology is homogeneous for the whole data sample and includes the derivation of H II region chemical abundances, structural parameters of bars and discs, galactocentric distances, and radial abundance profiles. We have obtained O/H and N/O abundance ratios from the Te-based (direct) method for a subsample of 610 regions, and from a variety of strong-line methods for the whole H II region sample. The strong-line methods have been evaluated in relation to the Te-based one from both a comparison of the derived O/H and N/O abundances for individual H II regions and a comparison of the abundance gradients derived from both methodologies. The median value and the standard deviation of the gradient distributions depend on the abundance method, and those based on the O3N2 indicator tend to flatten the steepest profiles, reducing the range of observed gradients. A detailed analysis and discussion of the derived O/H and N/O radial abundance gradients and y-intercepts for barred and unbarred galaxies is presented in the companion Paper II. The whole H II region catalogue including emission-line fluxes, positions, and derived abundances is made publicly available on the CDS VizieR facility, together with the radial abundance gradients for all galaxies.

Key words: ISM: abundances – H II regions – galaxies: spiral – galaxies: ISM – galaxies: abundances – galaxies: structure.

the present-day metallicity gradients have received much attention, 1 INTRODUCTION as they are the result of the galaxy evolution, where a complex The present-day distribution of metals in disc galaxies is a key di- interplay between star formation efficiency, infall of low-metallicity agnostic tool to understand their evolution. H II regions are excellent gas, metal-enriched gas outflows, interactions and mergers, stellar tracers of this distribution in the gas-phase of spirals. Although the migration, and gas flows within the disc shapes the radial distribution number of H II regions does vary from galaxy to galaxy, typically of metals. hundreds of regions populate galaxy discs, having high luminosities Radial gas flows cannot by themselves produce the radial metallic- concentrated in bright emission lines detectable up to large distances ity gradients observed in galaxies (Goetz & Koeppen 1992), but are (e.g. Rozas et al. 1999;Cedres´ & Cepa 2002), allowing us to map the known to be an efficient mechanism to change them (e.g. Schonrich¨ metal distribution across the face of galactic discs (e.g. Bresolin, & Binney 2009; Spitoni & Matteucci 2011; Grisoni, Spitoni & Garnett & Kennicutt 2004; Bresolin, Kennicutt & Ryan-Weber Matteucci 2018). The gravitational potential of non-axisymmetric 2012; Pilyugin, Grebel & Kniazev 2014). In the nearby Universe, structures in disc galaxies, such as bars, is one of the most efficient spectroscopic observations of H II regions revealed, to first order, an mechanisms that can create large-scale radial gas flows according to exponential decrease in the relative abundance of oxygen to hydrogen simulations (e.g. Sellwood & Wilkinson 1993; Athanassoula 2003). (normally used as a tracer for the metals) from the galaxy centre The induced gas flows mix the gas and can flatten the metallicity to the outer disc regions. The dependence of 12+log (O/H) with gradient: Assuming an initial negative metallicity profile, inwards galactocentric radius, normally termed radial metallicity profile, is gas flows dilute the higher metal content in the central regions, while then well parametrized by a straight line with a characteristic slope or outwards flows can enrich the more metal-poor outer disc areas. metallicity gradient. Since their discovery (Aller 1942;Searle1971), In the 90s, Vila-Costas & Edmunds (1992) were pioneering at investigating the properties of H II regions in relation to the properties of their host galaxies. In particular, they studied the gas-phase E-mail: [email protected] (AZ); [email protected] (EF); radial metallicity gradients. In this, and in subsequent work (Martin [email protected] (FB) &Roy1994; Zaritsky, Kennicutt & Huchra 1994; Dutil & Roy

C 2020 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society 2360 A. Zurita et al.

1999), a trend was reported for barred galaxies to show shallower the derived metallicities and, as a consequence, on the metallicity gra- metallicity gradients than unbarred galaxies, in agreement with the dients (Sanders et al. 2017; Zhang et al. 2017; Vale Asari et al. 2019). theoretical predictions. However, more recent work in which the Due to the most recent (mostly IFU-based) results, there might galaxy samples have been considerably enlarged, the comparison be a new general and extended conviction among astronomers that of gas-phase metallicity profiles of barred and unbarred galaxies the field is closed and bars have little impact on the gas-phase reveals no difference in the slope between the two types of spirals metallicity gradients. However, we believe there are enough reasons (Sanchez´ et al. 2014;Kaplanetal.2016;Perez-Montero´ et al. 2016; that justify a revision of the topic. Therefore, our aim is to benefit Sanchez-Menguiano´ et al. 2016; Zinchenko et al. 2019). from the improved spectroscopic data sets available nowadays to

Studies of the central gas-phase chemical abundances also yield make a compilation of emission-line measurements from resolved Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 contradictory results. For example, Ellison et al. (2011) found larger H II regions in nearby spirals. From these, we perform a homogeneous central metallicity in barred galaxies with respect to unbarred ones. analysis of the data and revisit the topic of the influence of stellar Considere` et al. (2000) and Dutil & Roy (1999) reported a lower bars on the gas-phase radial abundance gradient of O/H and N/O in central metallicity in a sample of barred starbursts than in unbarred spirals, aiming to solve the controversy raised by previous works. galaxies, while Cacho et al. (2014) reported no difference in barred We divide our study of the bar effect on the gas-phase radial abun- and unbarred galaxies. In a subsequent work, Florido et al. (2015) dance profiles into two parts. The first one is included in this paper, found no difference in central metallicity but an enhanced N/O ratio and it is organized as follows. Sections 2 and 3 present the galaxy and in the centres of barred galaxies with respect to unbarred galaxies H II region samples, respectively. The process to obtain physical prop- using Sloan Digital Sky Survey1 (SDSS, York et al. 2000)spectra, erties and chemical abundances for the H II regions with the electron as in Ellison et al. (2011) and Cacho et al. (2014). temperature-based (direct) and strong-line methods is described in Apart from these contradicting results, there are important differ- Section 4, where we also perform a comparison between the different ences in methodology between the cited works, especially between methods employed. In Section 5, we present the resulting radial the 90s works and the most recent ones. On the one hand, in the abundance profiles. Finally, in Section 6 we present a summary and 90s the chemical abundances were derived from integrated spectra our conclusions. The second part of the study is presented in the com- or spectrophotometry of individual H II regions, and authors used panion Paper II (Zurita et al. 2020), where we carefully analyse the different calibrations of the R23 parameter (Pagel et al. 1979)orthe derived radial metallicity gradients comparatively for (strongly and [O III]/Hβ and [N II]/[O III] line ratios to derive their metallicities. weakly) barred and unbarred galaxies. The results are compared with On the other hand, more recent works have preferentially used previous work, and discussed in the context of current knowledge on empirical calibrations of the O3N2 and N2O2 parameters (e.g. disc evolution and radial mixing induced by bars and spiral arms. Marino et al. 2013), the R calibration (Pilyugin & Grebel 2016), or other model-based methods such as the HII-CHI-mistry code (Perez-´ 2 GALAXY SAMPLE Montero 2014). Furthermore, the combination of spatial resolution of the Integral Field Unit (IFU) instruments used in these works, The galaxy sample is shown in Table 1. It comprises 51 nearby together with the median distance of the sample galaxies, implies spiral galaxies (distances <64 Mpc). Our criterion for the sample that these works are based either on a spaxel-by-spaxel analysis or selection was to include galaxies for which emission-line ratios of on integrated spectra of star-forming complexes, which normally H II regions were available from previous publications, together with include several individual H II regions.2 This is also the case for the the corresponding information on their celestial coordinates (either central abundances derived from SDSS spectra from Ellison et al. absolute or relative to the galactic centre). We concentrated the search (2011), Cacho et al. (2014), and Florido et al. (2015), which include on spirals with inclination angles of i < 70◦, with measurements for emission from the inner ∼1–4 kpc. at least seven H II regions covering a wide range in galactocentric It is well known that different gas-phase metallicities are obtained distance so that more reliable radial profiles of the chemical abun- depending on the strong-line method used (e.g. Kewley & Ellison dances can be obtained. As we want to minimize the amount of DIG 2008;Lopez-S´ anchez´ & Esteban 2010), but more importantly, this contamination on the derived H II region properties, we gave priority difference is not just a zero-point offset. Different methods also yield to galaxies with resolved spectroscopic data. Therefore, most of different radial metallicity gradients (Bresolin et al. 2009b; Arellano- the spectroscopic data of our sample comes from long-slit or fibre- Cordova´ et al. 2016). In the 90s, some of the largest data samples fed spectroscopic campaigns of individual H II regions. The median were based on inhomogeneous compilations of slopes derived from distance for the galaxy sample is 12 ± 2 Mpc, which yields an previous authors with different calibrations (Dutil & Roy 1999). In average spatial scale of 58 pc arcsec−1 (cf. ∼350 pc arcsec−1 for the more recent works, however, data analysis is homogeneous, but the CALIFA sample; Walcher et al. 2014). Typical extragalactic H II the improved instrumental sensitivity and the lower spatial resolution region diameters range from 90 to 800 pc (e.g. Rozas et al. 2000; imply contamination from the diffuse ionized gas (DIG; e.g. Zurita, Oey et al. 2003). Our search is rather exhaustive, but it is probably Rozas & Beckman 2000; Oey et al. 2007). The DIG has different not complete, and we cannot exclude that we have missed published physical conditions and ionizing spectra compared to H II regions, data that meet the above-mentioned requirements. The distribution and its inclusion on the extracted spectra can have a strong impact on of sample galaxy properties is presented in Section 2.3. There are other existing samples based on published H II region data, as it is the case for the compilation presented in Pilyugin et al. (2014). Our sample is considerably smaller (51 galaxies versus 130 1http://www.sdss.org in Pilyugin et al. 2014) because of our more restrictive criteria for the 2The spatial element-resolution in CALIFA is 3 arcsec, which corresponds to ∼1 kpc (Sanchez´ et al. 2014) at the average redshift of the survey. The spatial compilation: (1) We only include spiral galaxies in the sample, while resolution of VENGA is 5.6 arcsec, which corresponds to a median value of Pilyugin et al. (2014) also include irregulars; (2) we prioritized the ∼387 pc at the median distance of their sample galaxies (14.3 Mpc; Kaplan inclusion of data from resolved spectroscopy, but a considerable et al. 2016). Typical extragalactic H II region diameters range ∼90–800 pc number of galaxies in the Pilyugin et al. (2014) sample comes (e.g. Rozas, Zurita & Beckman 2000; Oey et al. 2003). from integral field spectroscopy (e.g. from Sanchez´ et al. 2012);

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2361

Table 1. Main properties of the galaxy sample: galaxy name (NGC number, except MW = Milky Way), morphological type, distance, absolute B-band magnitude, disc inclination (i), disc and bar position angles (PA, PAbar), disc isophotal (r25) and effective (re) radii, observed bar radius or semimajor axis (rbar), bar deprojected semimajor axis normalized to re, deprojected bar ellipticity (ebar,d), nuclear activity, the number of independent measurements of H II regions in each galaxy (in parenthesis those that allow Te-based abundance determinations).

a a a Galaxy Type DMB i PA PAbar r25 re rbar rbar,d/re ebar,d Nuc. N (NGC) (Mpc) (mag) (deg) (deg) (deg) (kpc) (kpc) (kpc) Act.b

MW .SBT4.. – −20.1 – – – 26.8c 4.5 ± 0.4d 5.0e 1.10 0.46f – 33 (32) 224g .SAS3.. 0.7 −21.0 77 38 55.7 20.6 9.4 ± 0.8 2.6 0.40 – Q 205 (22) Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 300 .SAS7.. 1.9 −17.9 56 ± 8 114 ± 8– 6.02.5± 0.3 0.0 – – – 42 (35) 598 .SAS6.. 0.8 −18.9 53 ± 822± 15 – 8.6 3.7 ± 0.3 0.0 – – – 316 (120) 628 .SAS5.. 9.0 −20.0 32 ± 34± 6 – 13.8 5.4 ± 0.9 0.0 – – – 210 (61) 925 .SXS7.. 9.3 −19.9 54 ± 6 120 ± 6 110 ± 3 14.2 8.3 ± 0.8 4.8 0.60 0.43 – 139 1058 .SAT5.. 10.6 −18.6 17 ± 14 150 ± 30 – 4.7 1.8 ± 0.4 0.0 – – S2 48 1068 RSAT3.. 10.1 −20.6 37 ± 4 175 ± 856± 4 10.4 2.4 ± 0.2 1.1 0.56 0.50 S1h 22 1097 .SBS3.. 16.0 −21.1 45 ± 5 117 ± 10 148 ± 5 21.7 8.8 ± 0.2 8.7 1.12 0.59 S3b 15 1232 .SXT5.. 14.5 −20.4 35 ± 2 104 ± 995± 5 15.7 8.2 ± 0.4 1.2 0.15 0.28 – 34 (13) 1313 .SBS7.. 4.6 −19.0 40 ± 61± 216± 16.12.8± 0.4 1.1 0.42 0.59 – 45 (8) 1365h .SBS3.. 19.6 −21.5 41 220 85 ± 1 32.0 11.4 ± 1.9 9.5 0.97 0.65 S1.8 87 (8) 1512i .SBR1.. 12.0 −19.4 49 ± 58045± 2 15.6 5.5 ± 0.8 4.1 0.92 0.64 – 147 (8) 1637 .SXT5.. 12.0 −19.1 39 ± 136± 276± 57.03.0± 0.1 1.4 0.53 0.50 – 17 1672 .SBS3.. 14.5 −20.6 34 ± 2 154 ± 799± 2 13.9 4.7 ± 0.2 5.1 1.24 0.72 S 17 2336 .SXR4.. 33.0 −22.0 56 ± 1 179 ± 5 122 ± 2 34.0 11.5 ± 1.3 4.6 0.67 0.58 – 36 2403 .SXS6.. 3.1 −19.0 57 ± 5 123 ± 4– 9.72.8± 0.1 0.0 – – – 77 (25) 2541 .SAS6.. 12.6 −18.9 60 ± 3 162 ± 5 171 ± 3 11.5 4.6 ± 0.2 1.6 0.36 0.26 – 19 2805j .SXT7.. 28.0 −21.1 36 ± 2 123 ± 3 114 ± 20 25.7 12 ± 2 2.1 0.17 0.11 – 17 2903 .SXT4.. 8.9 −20.6 59 ± 115± 324± 3 16.3 3.50 ± 0.09 3.1 0.93 0.57 – 53 (1) 2997 .SXT5.. 7.1 −19.9 48 ± 591± 7 109 ± 69.24.7± 0.6 1.1 0.26 0.23 – 22 (7) 3031 .SAS2.. 3.6 −20.4 54 ± 3 159 ± 2 – 14.2 4.5 ± 0.2 0.0 – – Q 131 (29) 3184 .SXT6.. 14.4 −20.4 16 ± 3 148 ± 10 82 ± 4 15.5 7.7 ± 1.1 1.9 0.25 0.20 – 84 (25) 3227 .SXS1P. 20.6 −20.4 66 ± 1 152 ± 3 160 ± 3 16.1 6.7 ± 0.7 1.8 0.35 0.27 S1.5 9 3310 .SXR4P. 18.7 −20.4 32 ± 5 172 ± 13 31 ± 20 8.4 4.9 ± 1.2 0.6 0.13 0.58 – 14 3319 .SBT6.. 14.1 −19.6 69 ± 226± 338± 1 12.6 8.8 ± 1.2 3.6 0.48 0.67 – 13 3344 RSXR4.. 9.8 −19.4 27 ± 5 150 ± 30 179 ± 1 10.1 3.50 ± 0.09 1.1 0.34 0.41 – 15 (1) 3351 .SBR3.. 10.5 −19.8 43 ± 212± 2 113 ± 2 11.3 4.7 ± 0.4 3.3 0.96 0.58 – 28 3359 .SBT5.. 16.7 −20.4 57 ± 2 176 ± 410± 1 17.6 7.0 ± 0.2 3.7 0.57 0.59 – 35 (3) 3521 .SXT4.. 12.1 −21.1 49 ± 6 159 ± 4 162 ± 2 19.3 4.5 ± 0.2 1.3 0.28 0.24 – 13 3621 .SAS7.. 7.2 −20.1 64 ± 4 163 ± 2 – 12.9 4.2 ± 0.4 0.0 – – – 80 (12) 4254 .SAS5.. 14.4 −20.7 39 ± 441± 10 – 11.3 4.8 ± 0.2 0.0 – – – 20 4258 .SXS4.. 8.4 −21.1 64 ± 1 159 ± 3 145 ± 5 22.8 5.6 ± 0.3 0.7 0.15 0.42 S2 65 (6) 4303 .SXT4.. 14.5 −20.7 30 ± 558± 20 2 ± 1 13.6 3.0 ± 0.6 2.5 0.93 0.63 S2 22 4321 .SXS4.. 17.2 −21.2 27 ± 331± 10 99 ± 4 18.6 5.9 ± 0.2 5.4 1.02 0.63 – 11 4395 .SAS9∗.4.3−17.6 55 ± 1 143 ± 7 120 ± 48.35.6± 1.0 2.9 0.60 0.74 S1.8 18 (5) 4625 .SXT9P. 9.5 −17.0 21 ± 6 160 ± 10 21 ± 63.00.9± 0.1 0.3 0.36 0.39 – 34 (1) 4651 .SAT5.. 29.1 −21.3 49 ± 279± 5 – 16.8 7.6 ± 0.40.0–––7 4654 .SXT6.. 14.5 −20.1 67 ± 4 127 ± 4 119 ± 2 10.3 7.5 ± 0.6 1.0 0.14 0.39 – 7 5194 .SAS4P. 8.0 −20.8 42 ± 838± 10 138 ± 3 13.0 7.7 ± 0.2 0.7 0.12 0.44 S2 104 (35) 5236 .SXS5.. 4.6 −20.3 16 ± 2 120 ± 40 55 ± 38.63.1± 0.1 2.7 0.90 0.63 – 94 (12) 5248 .SXT4.. 13.0 −19.9 58 ± 5 136 ± 795± 2 11.7 3.8 ± 0.4 1.4 0.58 0.54 – 11 5457 .SXT6.. 6.7 −20.9 36 ± 857± 30 79 ± 3 28.1 8.6 ± 0.2 1.6 0.19 0.43 – 260 (137) 6384 .SXR4.. 29.7 −21.8 46 ± 929± 436± 1 26.6 11.4 ± 1.9 3.5 0.31 0.37 – 18 6946 .SXT6.. 5.4 −20.9 29 ± 340± 15 14 ± 49.05.7± 0.4 1.8 0.33 0.57 – 10 (1) 7331 .SAS3.. 15.1 −21.5 63 ± 1 169 ± 2 – 23.0 8.0 ± 0.3 0.0 – – – 16 7518 RSXR1.. 34.8 −18.9 32 ± 641± 7 124 ± 2 7.2 3.20 ± 0.08 3.2 1.17 0.63 – 13 7529k RSA.4.. 63.2 −19.6 29 ± 5 157 ± 6– 7.33.3± 0.2 0.0 – – – 30 7591 .SB.4.. 53.2 −20.6 61 ± 2 142 ± 618± 3 15.0 6.2 ± 0.5 2.5 0.76 0.64 S 16 7678 .SXT5.. 44.7 −21.1 42 ± 325± 10 111 ± 10 16.5 6.8 ± 0.6 2.8 0.54 0.63 – 11 7793 .SAS7.. 3.6 −18.4 51 ± 399± 2– 4.92.3± 0.1 0.0 – – – 41 (3) a Notes. de Vaucouleurs et al. (1991, RC3). MB obtained from the extinction-corrected total blue magnitude and the distances in column 3, except for the Milky Way (Licquia, Newman & Brinchmann 2015). bNuclear activity according to Veron-Cetty´ & Veron´ (2006). c From Goodwin, Gribbin & Hendry (1998), r25 = (26.8 ± 1.1) kpc. d From the disc scale length reported by Licquia, Newman & Bershady (2016), rd = (2.71 ± 0.22) kpc. e From Wegg, Gerhard & Portail (2015), rbar = (5.0 ± 0.2) kpc. fFrom Martinez-Valpuesta & Gerhard (2011). g Disc inclination and PA from Corbelli et al. (2010), re from reported value for the disc scale length in the I band by Courteau et al. (2011), adjusted to D = 744 kpc, rd = (5.6 ± 0.5) kpc. Bar parameters from Blana˜ D´ıaz et al. (2017, 2018). h Disc inclination, PA, and rd from Zanmar´ Sanchez´ et al. (2008). iDisc PA from Koribalski & Lopez-S´ anchez´ (2009). jDisc inclination and PA from Erroz-Ferrer et al. (2015). kMorphological type from the Hyperleda data base (Makarov et al. 2014, http://leda.univ-lyon1.fr/).

MNRAS 500, 2359–2379 (2021) 2362 A. Zurita et al. Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 1. (Left) Distribution of deprojected bar ellipticities (ebar,d) for the three different RC3 visual bar classes: unbarred, SA galaxies (dashed blue), barred, SB (dashed orange), and intermediate or SAB (dotted green). (Right) Deprojected bar ellipticity as a function of the normalized (to the disc effective radius) bar radius. The straight dotted line marks the division between barred and unbarred galaxies. The grey dots represent bar parameters for the face-on barred galaxies analysed by Gadotti (2009) for comparison.

(3) we only compiled data for which both emission-line fluxes and The morphological characterization of the galaxies was performed H II region coordinates were available, in order to recalculate both through the widely used method of fitting ellipses to the image abundances and galactocentric distances with the same methodology, isophotes (e.g. Wozniak et al. 1995; Aguerri et al. 2000;Marinova& and (4) our minimum required number of regions per galaxy was set Jogee 2007) with the ellipse task within IRAF.4 For each galaxy, to seven, while according to fig. 1 in Pilyugin et al. (2014) this limit is the position of the nucleus was previously determined and fixed smaller in their sample (∼3 regions/galaxy). The latter requirement, during the fit, whereas the PA and the ellipticity (e) of the isophotal together with the fact that our sample benefits from the publications ellipses were set to be free parameters. on resolved H II spectroscopy of the last few years, implies that we The disc scale length (rd) was calculated from a fitting, of the have a larger average number of H II regions per galaxy, ∼56 against disc-dominated area of the surface brightness radial profile obtained ∼29 in Pilyugin et al. (2014). Another advantage of our sample is with ellipse, to a single exponential function. When a break the availability of Te-based abundance estimates, as we compiled in the surface brightness profile (e.g. Munoz-Mateos˜ et al. 2013) auroral-line fluxes in addition to strong-line fluxes (see Section 3). was detected or we suspected could be present (e.g. for NGC 1058, This work requires knowledge of several galaxy structural param- NGC 3521, NGC 4258, and NGC 4303), we derived the disc scale eters: (1) The disc inclination and position angle (PA) are necessary length from the innermost part of the disc-dominated profile by to compute H II region deprojected galactocentric distances. The disc excluding the outer disc region from the fit. Therefore, our derived scale length (or equivalently the disc effective radius) is also needed rd values correspond to the inner disc scale length in these cases. The 5 to normalize distances and sizes. (2) Bar parameters (length, PA, and disc effective radius, re, was calculated from rd. The disc PA and ellipticity) are also required. These allow us to quantify bar strengths ellipticity (and therefore the disc inclination i) were estimated from and to determine H II region positions in relation to the bar. the average values in the disc-dominated region of the corresponding In spite of the fact that the galaxies of the sample are nearby and radial profiles. Table 1 shows the disc parameters obtained from well-known targets, when structural parameters are available neither our photometric analysis, except for the Milky Way6 and M31 for the methodology nor the photometric bands employed by different which we used published values. For NGC 1365, NGC 1512, and authors are homogeneous. In order to ensure consistency in the NGC 2805, we have also partially used morphological parameters data and in the corresponding analysis procedure, we recalculated from the literature as compiled images were not suitable for the disc and bar structural parameters in a homogeneous way for the whole analysis. See Table 1 for the references. sample galaxies. The methodology and results are described in the following sections. 2.2 Bar parameters and classification The bar parameters (length, PA, and ellipticity)7 were also deter- 2.1 Morphological analysis mined from the radial profiles obtained from the ellipse fits. The bar We compiled broad-band images for all galaxies in the sample radius was defined as the radius at which the ellipticity profile shows (except for M31 and the Milky Way). All images are Sloan r-band or a local maximum value, ebar, whereas the PA remains approximately Johnson R-band images, except for NGC 1637 and NGC 1365, for constant. This method has the advantage that it is simple to reproduce whichweusedaV-band and I-band image, respectively. Most of the images come from the SDSS data base, although a relevant number of 4 them were obtained from the NASA/IPAC Extragalactic Database,3 IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy and two galaxies (NGC 2336 and NGC 6384) were observed with (AURA) under a cooperative agreement with the National Science Founda- CAFOS at the Calar Alto Observatory 2.2 m telescope in 2018 August tion. through an r-band filter. Table B1 (available online) summarizes the 5The disc effective radius is the radius that contains half of the total disc origin and photometric bands for these images. integrated flux. For an exponential disc profile, re = 1.678rd, with rd being the disc scale length. 6See brief explanation on the choice of parameters for the Milky Way in Appendix D (available online). 3 7 https://ned.ipac.caltech.edu/ The bar ellipticity, ebar, is related to the bar semi-axes by ebar = 1 − b/a.

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2363

Nair & Abraham 2010;Erwin2018). The classification has a certain degree of ambiguity. In fact, several galaxies classified as SA in RC3 (NGC 2541, NGC 1068, NGC 5194, NGC 4395, NGC 224) contain a central oval distortion (Scoville et al. 1988; Zaritsky, Rix &Rieke1993; Chapelon, Contini & Davoust 1999; Buta, Corwin & Odewahn 2007; Beaton et al. 2007), while other galaxies classified as SAB (e.g. NGC 2903 and NGC 925) have bars comparable in length and/or ellipticity to those in SB galaxies or have no

apparent bar, as is the case for NGC 2403. This can be better Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 seen in Fig. 1 (left) (also in fig. 15 of Marinova & Jogee 2007, for a different sample), where we plot for our sample galaxies the distribution of our estimated deprojected bar ellipticities (ebar,d) for the three different bar classes separately, as recorded from the RC3 catalogue. It can be seen that SB galaxies (in orange) are all concentrated in the largest ellipticity range (as expected), with ebar,d ≥ 0.4, but SAB galaxies (in green) have significant overlap in ellipticity with SB galaxies and cover a wide range in ebar,d from 0.2 to around 0.7, and the four galaxies classified as SA (in blue), but having central oval distortions, have ellipticity values of up to ∼0.7. The deprojected bar ellipticity is frequently used to quantify the bar strength (e.g. Martin 1995; Abraham & Merrifield 2000), Figure 2. Histograms showing the distribution of absolute integrated B- and correlates with the bar gravitational torque (Q –D´ıaz-Garc´ıa band magnitude, morphological T-type, disc inclination angle, and disc b et al. 2016). Therefore, in order to perform a more quantitative effective radius (re) for the strongly barred (dotted orange), weakly barred (dashed green), and unbarred (blue) subsamples of galaxies separately. The classification of bars, we have used the deprojected bar ellipticity to ≥ distribution for all galaxies is shown with a grey dashed line. The circles in classify barred galaxies into strongly barred (when ebar,d 0.5) and the upper side of the panels indicate the median value for each distribution weakly barred (when 0.3 ≤ ebar,d < 0.5). Any galaxy with measured and the horizontal error bar covers the 95 per cent confidence interval for the central oval distortion in the fitted isophotes, but for which the corresponding data value. The number of galaxies in each subsample and the deprojected ellipticity is smaller than 0.3 was considered as unbarred. P-value for the two-sample AD test (AD in per cent) for the distribution of A similar criterion was also adopted by several authors (see e.g. strongly barred and unbarred galaxies are also shown. The AD P-values are Martin 1995; Abraham et al. 1999; Marinova & Jogee 2007). higher than 5 per cent in all cases, indicating that the distributions are not There is a relation between bar length and ellipticity, in the sense different in strongly barred and unbarred galaxies. that very elliptical bars are also long. And for a given bar length, there is a minimum observed ellipticity. This can be seen in Fig. 1 and to compare with published values in the literature, but see Erwin (right). Our cut-off in ellipticity implies that we are considering as (2005) for a discussion on different methods to estimate the bar length unbarred, galaxies with detected central oval structures of axial ratio 8 and their systematic differences. In particular, the radius obtained >0.7 and with a radius of 2.3 kpc (or 0.4 re) . from this procedure must be considered as a lower bound to the In summary, after the bar classification described above, our visual bar length, although the difference is not as large as initially sample comprises 22 strongly barred, 9 weakly barred, and 20 thought (D´ıaz-Garc´ıa et al. 2016). The bar PA (PAbar) was estimated unbarred galaxies, i.e. there is virtually equal representation of from the average PA within the bar radius. strongly barred and unbarred systems. Any oval feature with an observed ellipticity above ∼0.2 was initially considered as a potential barred structure. The observed 2.3 Distribution of galaxy properties values for the bar length and ellipticity were afterwards depro- jected to their face-on values, using the analytical expressions in Fig. 2 shows the distribution of absolute integrated B-band mag- Gadotti et al. (2007) (in their appendix A), considering the bar as nitude, morphological T-type, inclination angle, and disc effective a planar ellipse. Although this picture is a simplification of the radius, separately for the unbarred, weakly barred, and strongly real 3D shape of bars, the typical intrinsic flattening (∼0.34 on barred galaxy subsamples, obtained as explained in Section 2.2. average – Mendez-Abreu´ et al. 2018) is similar to that of stellar Median values are similar for all subsamples within errors. We have discs. In addition, Zou, Shen & Li (2014) used mock galaxies to performed a two-sample Anderson–Darling (hereinafter AD) test quantify uncertainties in bar deprojected parameters and conclude to the distributions of strongly barred and unbarred galaxies in all that the 2D deprojection method used here is preferred over 1D four parameters. The AD P-values9 are higher than 5 per cent in all methods, as it yields reasonably good uncertainties (∼10 per cent for galaxy inclination angles of 60◦) in both bar length and 8 ellipticity. M31 has also been considered as unbarred in this work. It has a boxy/peanut- ∼ Disc galaxies of a given morphological type are generally classi- shaped bulge with an inner thin bar extending to 0.4 re (Beaton et al. 2007, Blana˜ D´ıaz et al. 2017, 2018). Its ellipticity is not available, but modelled fied into classes according to their bar visual prominence. Following isophotes (Blana˜ D´ıaz et al. 2018) indicate a low bar ellipticity (∼0.3) that, the RC3 catalogue (de Vaucouleurs et al. 1991) nomenclature, the with our criteria, would correspond to an unbarred (or maybe weakly barred) standard classes are SB for strong bars, SA for unbarred galaxies, galaxy. We want to note that considering M31 as a weakly barred galaxy and SAB for intermediate or weak bars. However, in this respect would not modify our results. the RC3 bar-type classification is known to be problematic and 9The output of the AD test is a P-value or significance level at which the should be taken with caution (see e.g. Marinova & Jogee 2007; null hypothesis can be rejected. A threshold level of ∼5 per cent is usually

MNRAS 500, 2359–2379 (2021) 2364 A. Zurita et al. cases, indicating that the distributions are not significantly different few cases such as the M31 data by Sanders et al. (2012), and those in strongly barred and unbarred galaxies. for NGC 1672, NGC 5248, and NGC 1097 by Storchi-Bergmann, Our galaxy sample is not intended to be complete in any parameter, Wilson & Baldwin (1996), only observed fluxes were given by the but it is important to ensure that barred and unbarred galaxies cover authors. In these particular cases, we dereddened the lines using a similar parameter space. The comparison of Fig. 2 confirms that the Cardelli, Clayton & Mathis (1989) extinction curve and either this is the case. the Av values reported by the authors or the observed Hα/Hβ ratio by imposing a theoretical Balmer decrement Hα/Hβ = 2.86 (Osterbrock & Ferland 2006). 3HII REGION SAMPLE (v) We tried to be conservative and excluded from our compilation Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 Our observational data come from a compilation of optical emission- objects whose data were identified as uncertain by the authors in line fluxes and positions of H II regions from published papers. We their original papers or were likely to be uncertain (because no then used a consistent method to obtain metallicities and deprojected determination of the extinction coefficient was done and/or very galactocentric distances for all the regions. few lines were detected). In addition, some H II region data were Our criteria for the sample selection were described in Section 2. discarded during the analysis process due to unphysical results that For these galaxies, we collected H II region emission-line fluxes could be indicative of errors. In Appendix D, available online, we and the corresponding information on their position within the host include specific notes for some of the galaxies. galaxies. It is a major compilation of published data from over The final sample contains 2831 independent measurements of H II 80 different papers. See Table A1 (available online) for a list of regions belonging to 51 nearby spirals (including the Milky Way). references. For 709 of them, auroral-line fluxes for the determination of the Te-based oxygen abundance are available. 3.1 H II region emission-line fluxes

The compilation includes fluxes, normalized to those of the 3.2 H II region location Hβ line, for the brightest emission lines ([O II] λλ3726, 3729,10 [O III] λλ4959, 5007, [N II] λλ6548, 6583, Hα,[SII] λλ6717, 6731, We have recalculated the deprojected radial positions for all the H II [S III] λλ9069, 9532). The auroral-line fluxes ([S II] λλ4068, 4076, regions from the celestial coordinates, from offset positions from the [O III] λ4363, [N II] λ5755, [S III] λ6312, and [O II] λλ7320, 7330) galaxy centre or from distances along a slit of a given centre and were also compiled when available. These are necessary for a more PA, depending on the information given by the different authors in the original publications (see Table A1). We have assumed thin discs accurate determination of chemical abundances via the Te-based or direct method. and have employed the disc PAs and inclinations derived from our The original data are highly heterogeneous in a number of aspects, morphological analysis of the galaxies (or from the literature for a in particular in the data format for both positions and fluxes, in the limited number of galaxies; see Table 1 and Section 2.1). Our derived number of emission lines provided by the different authors, or in the values for the disc effective radius (re) and the isophotal galaxy radius flux error estimates. We adopted the following criteria: (r25, from RC3), given in Table 1, have been used to normalize the derived H II region deprojected galactocentric distances. (i) When either only the brightest line of the [O III] λλ4959, 5007, [N II] λλ6548, 6583, or [S III] λλ9069, 9532 doublets is given or the sum of the two line fluxes is given, we calculated the flux of individual 3.3 AGN contamination and circumnuclear regions = lines assuming the theoretical ratios: [O III] λ5007/[O III] λ4959 3, Nuclear activity can potentially contaminate the spectra of the II II = III III = [N ] λ6583/[N ] λ6548 3, and [S ] λ9532/[S ] λ9069 2.44 innermost H II regions in active galactic nuclei (AGNs), which can (e.g. Perez-Montero´ 2017). therefore produce an ill determination of chemical abundances. In II (ii) The compilation contains observations of the same H re- particular, enhanced [N II]/Hα from AGNs could be misinterpreted gions by different authors. We retained all of them as independent as an enhanced nitrogen abundance. Our galaxy sample contains 13 observations. galaxies with nuclear activity (1 LINER, 10 Seyferts, and 2 QSOs), (iii) Emission-line flux uncertainties were not always given by according to Veron-Cetty´ & Veron´ (2006). See column 13 in Table 1. the authors. We used the measurements of different authors for Fig. 3 shows the BPT (Baldwin, Phillips & Terlevich 1981) II thesameH region to estimate errors in the compiled emission diagram for all the H II regions of the sample. As we can see, the ∼ lines. Typical differences between different authors are 15 per cent position of the H II regions is compatible with photoionization by ∼ for [O III] λλ4959, 5007 and [N II] λλ6548, 6583, 20 per cent for massive stars in most cases. Only a small percentage of regions is [S II] λλ6717, 6731, and around 30 per cent for [O II] λ3727 and above the demarcation line by Kewley et al. (2001). Most of these [S III] λλ9069, 9532. For the auroral-line fluxes, there was more regions are still compatible with photoionization if we take into uniformity among authors and we have compiled their reported account observational errors, and the majority belongs to normal uncertainties. (non-AGN) galaxies. (iv) In most cases, the emission-line fluxes published by the In any case, to be on the safe side, we have been scrupulous with different authors had already been corrected for internal extinction. the data and have ignored in our radial abundance profile fits the In these cases, the provided corrected fluxes have been used. In a innermost regions (1.5 kpc) of active galaxies. This is the case for NGC 1058, NGC 1365, NGC 1672, NGC 3227, NGC 4395, adopted. Therefore, values below 5 per cent are statistically significant and NGC 4395, and NGC 5194. H II regions simultaneously above the indicate that the null hypothesis that both samples are drawn from the same demarcation line by Kauffmann et al. (2003)inthe[OIII] λ5007/Hβ parent distribution can be rejected. versus [N II] λ6583/Hβ diagram (left-hand panel in Fig. 3)and 10The [O II] λλ3726, 3729 is usually blended for the compiled observational the one by Kewley et al. (2001)inthe[OIII] λ5007/Hβ versus data. Hereinafter, it will be referred to as [O II] λ3727. [S II] λλ6717, 6731/Hβ diagram (right), as well as circumnuclear

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2365 Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 3. [O III] λ5007/Hβ versus [N II] λ6583/Hα (left) and [O III] λ5007/Hβ versus [S II] λ6717, 6731/Hα (right) BPT diagrams for the H II regions compiled for this work. The solid green and dashed black lines separate the region for star-forming galaxies from the region for AGNs according to Kewley et al. (2001) and Kauffmann et al. (2003), respectively. The dotted–dashed blue line shows the fit obtained by Bresolin et al. (2012) to the sequence outlined by a sample of Galactic and extragalactic H II regions. The dotted-black and dotted-green lines enclose the area in which regions located over the Kauffmann et al. (2003) and Kewley et al. (2001) dividing lines, respectively, would still be compatible with star forming, assuming an observational error of ∼15 per cent for [N II]and[OIII], and 20 per cent for [S II]. H II regions located in weakly barred, strongly barred, and unbarred galaxies are shown in green, orange, and blue, respectively. Overplotted black squares mark H II regions located in galaxies containing an AGN at distances ≤1.5 kpc from the nucleus. Overplotted red dots mark circumnuclear H II regions. The black dots mark H II regions that were excluded from the fittings for being simultaneously above the demarcation line by Kauffmann et al. (2003)inthe[OIII] λ5007/Hβ versus [N II] λ6583/Hα diagram and the one by Kewley et al. (2001)inthe[OIII] λ5007/Hβ versus [S II] λλ6717, 6731/Hα diagram.

H II regions or hotspots (e.g. Kennicutt, Keel & Blaha 1989) in non- the most accurate method for H II regions in nearby galaxies. The AGN galaxies (NGC 2903, NGC 2997, NGC 3351, NGC 4321, methodology to obtain the electron temperature and density, and NGC 5236, NGC 5248, 52 regions in total), will not be considered Te-based and strong-line chemical abundances is described in the for the determination of radial abundance gradients. We have opted, following sections. however, to calculate the corresponding strong-line abundances for these regions, in order to assess the behaviour of the different methods for these objects. 4.1 Electron density and temperature We find a total of 89 regions that are either innermost H II regions in AGNs, hotspots, or nebulae displaying signatures of shock excitation The physical conditions of the H II regions have been derived from in the BPT diagrams. Five of them have auroral-line measurements. the corresponding collisionally excited lines and the task temden We will further comment on this in Section 5. within the IRAF nebular package (Shaw & Dufour 1995), with updated values for the atomic parameters (collisional strengths and transition probabilities) as shown in table 5 of Bresolin et al. (2009b). Our compilation of emission-line fluxes includes the auroral lines 4 PHYSICAL PROPERTIES AND CHEMICAL [S II] λλ4068, 4076, [O III] λ4363, [N II] λ5755, [S III] λ6312, and ABUNDANCES [O II] λλ7320, 7330 (referred to as [O II] λ7325 in the following), The determination of the electron temperature, Te, is a necessary when reported by the different authors. In total, the sample comprises step to obtain accurate chemical abundances in H II regions via 709 H II regions with flux measurements for at least one of the auroral the so-called Te-based or direct method. This is normally done by lines, of which we could derive the electron temperature (from at least measuring a temperature-sensitive emission-line ratio (auroral-to- one auroral line) for 639 H II regions. nebular), which requires the detection of the faint auroral lines, Density and temperature have been obtained simultaneously in an together with the corresponding nebular bright lines for the same iterative process. The electron temperature (Te) was initially set to ion [e.g. [O III] λ4363/[O III] λλ4959, 5007 for the determination of 7500 K, and a first value for the electron density, ne, was obtained Te([O III])]. from the [S II] λ6717/[S II] λ6731 line ratio. This initial value for ne Our compilation includes auroral lines for a limited number of H II is then used for a preliminary determination of the electron temper- regions (709 out of 2831 regions). Therefore, Te measurements and ature and the ionic abundances (as described in the next section). Te-based abundances can only be obtained for about 25 per cent of the Using these initial estimates of the electron temperatures and ionic regions (and reliably for just ∼20 per cent). For the rest of the H II re- abundances, the electron density is recalculated, and the reddening- gions, only estimates based on strong emission lines can be obtained. corrected fluxes of [N II] λ5755 and [O II] λ7325 are corrected for For the regions with auroral-line detection, we have recalculated Te- recombination contamination following Liu et al. (2000), where we 2+ + 2+ + based abundances with a homogeneous methodology. This allows have assumed N /N = O /O . Regions with Te above the range us to evaluate the different strong-line abundances in relation to of validity of Liu et al. (2000) relations have not been corrected.

MNRAS 500, 2359–2379 (2021) 2366 A. Zurita et al.

Recombination corrections are small, and median values for the corrections are 3.0 and 0.6 per cent for [O II] λ7325 and [N II] λ5755, respectively. The ionic and total abundances were then re-calculated and the iterative process was then repeated until convergence was achieved. The electron density has been determined for 1419 H II regions. The vast majority (92 per cent) of them have electron densities below − ∼200 cm 3. For those regions with no reported [S II] λλ6717,6731

fluxes or with [S II] λ6717/[S II] λ6731 ratios above the low density Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 limit, we have adopted an electron density of 29.2 cm−3 (the median valueforthesampleHII regions). This particular choice of ne has virtually no effect on the subsequent temperature and abundance Figure 4. Comparison between our derived Te-based O/H (left) and N/O estimates, since we remain within the low-density regime. (right) abundances, in the y-axis, with those obtained by other authors in All five values of the electron temperature Te[O III], Te[O II], previous studies (x-axis). The red dashed lines mark the 1:1 relation. Data Te[N II], Te[S II], and Te[S III]havebeenmeasuredinonly52HII points in light blue mark the regions with larger deviation from the 1:1 relation. regions, but in 377 H II regions we derived at least two values of Te, and at least one measurement of Te was obtained for 639 H II abundances, as this temperature yielded abnormal abundances in regions. In the calculation of the electron temperature, we have comparison with other H II region abundances in the same galaxy. discarded values with resulting uncertainties above 40 per cent This is the case for only 14 H II regions: 3 in NGC 300, 2 from Pagel and/or absolute errors larger than 4000 K. The absolute errors et al. (1979) and 1 region from Edmunds & Pagel (1984), and 11 in for Te have been estimated from error propagation of the relevant NGC 3184 from Berg et al. (2020). line flux uncertainties. These are in the range 100–3800 K for all The final adopted values for the low- and high-ionization zones determinations, with median values of ∼400–600 K for the five are available in the corresponding electronic table available on CDS determinations of Te. VizieR (via https://vizier.u-strasbg.fr/viz-bin/VizieR). Fig. C1 (available online) shows the relations obtained for electron The task ionic within the IRAF nebular package was + + temperatures for the different ionic species. The derived values of Te employed for obtaining the ionic abundances of N ,O,and support the relations predicted by photoionization models (Stasinska´ O2+. The task assumes a five-level atom approximation and yields 1982;Garnett1992). This is especially remarkable for the relation the ionic abundance for the given zonal temperature, electron between Te[S III]andTe[O III], and Te[S III]–Te[N II]. There is wide density, and corresponding reddening-corrected emission-line ratios + observational evidence supporting these theoretical relations (e.g. ([N II] λ6584/Hβ,[OII] λ3727/Hβ,and[OIII] λ5007/Hβ,forN , Bresolin et al. 2009b; Croxall et al. 2015;Perez-Montero´ 2017;Berg O+,andO2+, respectively). Absolute errors in the ionic abundances et al. 2020) but based on smaller data samples. have been estimated from the propagation of the error in the adopted electron temperature. We have been very conservative with our 4.2 T -based ionic and total abundances calculations. In cases where the authors give auroral-line fluxes that e are afterwards not used for temperature and/or abundance estimates, + + + The N ,O ,andO2 ionic abundances have been calculated from the we have chosen not to use them, whether or not this is justified by fluxes of collisionally excited emission lines, assuming a two-zone the authors (e.g. Croxall et al. 2015, 2016; Lin et al. 2017). scheme for the ionization structure of the H II regions, with two zones The total oxygen abundance is then obtained from the sum of + + 2+ + 3+ + of different Te where atomic species of similar ionization potential O /H and O /H . We therefore assume that the amount of O /H coexist. In view of the relations for Te derived above for the different is negligible, which might not be true for high-excitation H II regions. 3+ atomic species, we have adopted Te([N II]) as the temperature for Photoionization models predict a fraction of O /O 1 per cent only + + + + 2+ the low-excitation region (N ,O ), and Te([O III]) as the electron in the highest excitation regions, those for which O /(O + O ) temperature for the highest ionization zone. The following criteria 10 per cent (Izotov et al. 2006). Only five H II regions11 have low have been adopted when these temperatures were not available: O+/(O+ + O2+), with values in the range 7–9 per cent. The high- excitation line He II λ4686 is only detected in NGC 598 MA1 by (i) When T [N II] could not be measured, the temperature for e Kehrig et al. (2011), and these authors estimate O3+ contributions the low-excitation zone has been obtained from the average value that would increase 12+log (O/H) by ∼0.06 dex, which is smaller obtained from T [O III]andT [S III] and the inversion of the cor- e e than our quoted uncertainty (0.19 dex, with an average of 0.12 dex responding Garnett (1992) relations, or from just one of them, for all regions). We can therefore safely assume O/H = O+/H+ + depending on availability. + + O2 /H for all H II regions. (ii) When T [O III] could not be measured, we have obtained the e For nitrogen, we assume the usual relation N/O = N+/O+ high-ionization zone temperature from either T [S III]orT [N II]and e e (Peimbert & Costero 1969), based on the similarity of the O and the inversion of the Garnett (1992) relations, or from the average N ionization potential. of the two resulting estimates of T [O III], when both T [S III]and e e Fig. 4 shows a comparison of our derived total oxygen (left) and T [N II] were available. e nitrogen abundances (right) with those calculated by the original Our measured values for Te[S II] and/or Te[O II] have only been authors (Table A1, available online). The agreement is good over the employed when no other electron temperature was available. For whole range of abundances, with a scatter of 0.12 and 0.09 dex rms these cases, either Te[O II] or the average of both Te[S II]andTe[O II] for 12+log (O/H) and log (N/O), respectively. Points deviating more (when both were available) have been used as representative for the low-excitation zone, and the high-excitation zone temperature has been obtained from the inversion of equation (1) of Garnett 11M101 N5451C, NGC 598 MA1, NGC 598 IC132, MW NGC 3603, and (1992). When only Te[S II] was available, we have not reported ionic MW Sh2-83

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2367 than three times the scatter from the 1:1 relation have been marked in N2S2 (=log ([N II] λ6583/[S II] λλ6717, 6731)) provided by Perez-´ the plots with light blue symbols. These constitute a small percentage Montero & Contini (2009). of the total sample: 21 regions or 3.7 per cent of the total, and 8 regions or 4.7 per cent for O/H and N/O, respectively. These outliers 4.4 Comparison between T - and strong-line-based abundances correspond to data collected from different authors, but ∼60 per cent e comes from the work by Magrini et al. (2010) and Stanghellini et al. We have employed the O/H and N/O abundance ratios obtained (2010), which suggests differences in the methodology followed by from our direct estimate of Te (Section 4.2) to check the reliability us and by these authors in the derivation of the total abundances. of the strong-line methods used in this work (described in the

previous section). There are several detailed evaluations of these Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 methods in the literature, which either compare T -based abundances 4.3 Strong-line abundance estimates e with empirical calibrations (e.g. Bresolin et al. 2004, 2009b;Perez-´ Direct or Te-based abundances are only available for about 20 per cent Montero & D´ıaz 2005; Yin et al. 2007;Lopez-S´ anchez´ & Esteban of the H II regions (610 out of 2831). We have then considered 2010; Pilyugin & Grebel 2016) or different strong-line methods a number of different abundance determination methods based on among themselves (e.g. Kewley & Ellison 2008; Rupke, Kewley strong-line flux ratios to estimate 12+log (O/H) and log (N/O). As &Chien2010;Hoetal.2015). we are interested in the galactic radial profiles of the chemical Our purpose here is to test the performance of the methods abundances, a comparison is necessary, as there are well-known employed in this work in relation with the direct method for our offsets between the different methods, but more importantly, different specific sample of galactic and extragalactic H II regions. The aim methods can also yield different radial gradient slopes (e.g. Bresolin is to study possible systematic trends that could affect the radial et al. 2009b; Moustakas et al. 2010; Arellano-Cordova´ et al. 2016). abundance gradients, and that may be helpful at explaining our and The methods used in this work have been selected among many previous (discrepant) results. others because they are some of the most frequently used, and/or The left-hand panel of Fig. 5 shows a comparison between the because they allow us to compare with previous studies: different strong-line metallicities and the Te-based ones. We have opted to show the comparison in the form of 2D density plots as, given (i) The HII-CHI-mistry method (hereinafter HCM, version 3.0) the high overlapping in data points, scatter plots are more difficult to makes use of grids of photoionization models by Perez-Montero´ analyse. The same colour scale has been used for the seven panels, (2014) to calculate 12+log (O/H), in addition to the N/O abundance where darker areas correspond to a higher concentration of data ratio. points. The number of regions used for each comparison is shown (ii) O3N2 = log [([O III] λ5007/Hβ)/([N II] λ6583/Hα)] as cali- in the upper left corner of each panel. This number is generally brated by Pettini & Pagel (2004), and later by Marino et al. (2013), smaller than the total number of regions with T -based abundances hereinafter PP04 and M13, respectively. e (610) because for some of these regions either the relevant line (iii) N2 = log ([N II] λ6583/Hα) with the empirical calibration ratios are not available or these are outside the validity range of the given by Pettini & Pagel (2004). strong-line calibrations. The latter is the reason why the comparison (iv) N2O2 = log ([N II] λ6583/[O II] λ3727) with the empirical with T -based abundances for the two calibrations of the O3N2 calibration of Bresolin (2007), hereinafter B07. e parameter involves a different number of regions. A first look at the (v) R = ([O II] λ3727 + [O III] λλ4959, 5007)/Hβ, as calibrated 23 plots shows the expected behaviour: a positive correlation between from theoretical model grids by McGaugh (1991), hereinafter MG91, strong-line abundances and T -based ones (Spearman’s coefficient, with the Kobulnicky, Kennicutt & Pizagno (1999) parametrization e ρ>0.5 with all methods), and the well-known overestimation of (hereinafter KKP99). The well-known disadvantage of this indicator the oxygen abundances obtained from R (bottom-right panel) by is that it is double valued, i.e. the same value of R can yield 23 23 ∼0.3 dex (e.g. Kewley & Ellison 2008; Bresolin et al. 2009b), as two different values for 12+log (O/H), making it necessary to use we have used a calibration from photoionization models (McGaugh secondary indicators to break the degeneracy. Initially, we tried the 1991). The median value of the difference in 12+log (O/H) between prescription of other authors to use the ratio [N II] λ6583/[O II] λ3727, the strong-line methods and the T -based abundances, which we by assigning the upper branch when this ratio is larger than −1.2 e will term residuals, are in all cases (except for R ) smaller than (e.g. Kewley & Ellison 2008). However, this produces artificial 23 the dispersion (rms deviation from equality), with the latter being discontinuities in some of the derived radial abundance profiles (e.g. pretty similar, ∼0.22–0.25 dex, in all cases. However, important for NGC 2403, NGC 300, NGC 925, NGC 925). In addition, the differences between methods arise when we analyse the residuals as a [N II] λ6583 emission-line flux was not available for a considerable function of 12+log (O/H)Te , log (N/O)Te ,andO32(=[O III] λλ5007, number of H II regions. After a careful analysis of R23 as a function 4959/[O II] λ3727), the latter being a proxy for the ionization pa- of other oxygen abundance indicators, and diagnostic emission- rameter (right-hand panels of Figs 5 and 6). In this respect, the line ratios, we found it more reliable to use our derived HCM strong negative correlation (ρ =−0.77) of the residuals of the oxygen abundances for breaking the R degeneracy: regions with 23 O3N2 method as calibrated by M13 with the T -based oxygen 12+log (O/H)<7.9 as estimated with HCM were assigned to the e abundances is remarkable. This implies that oxygen abundances lower branch, and all the rest to the upper branch. For 2458 H II in the higher metallicity range are underestimated with respect to regions with R measurements, 48 (2410) were assigned to the 23 T -based abundances by up to ∼0.3–0.4 dex, while the opposite lower (upper) branch. e occurs for the lower metallicity regions. That translates in a much (vi) The R calibration by Pilyugin & Grebel (2016), here- lower range in 12+log (O/H) for H II region oxygen abundances inafter PG16, that uses three bright emission-line ratios (R2 = obtained with this method, with respect to the range covered by their [O II]λ3727/Hβ,R3= [O III] λλ4959, 5007/Hβ,andN2= corresponding Te-based abundances. This was already visible in the [N II] λλ6548, 6583/Hβ) to determine 12+log (O/H) and log (N/H). corresponding left-hand panel in Fig. 5, where we can see that O3N2 For the N/O abundance ratio, in addition to HCM and the R abundances are all in the range 8.2–8.7 dex, while the Te-based ones calibration, we have also used the empirical calibrations of N2O2 and cover a wider range 7.7–8.9. We should recall here that we have used

MNRAS 500, 2359–2379 (2021) 2368 A. Zurita et al. Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 5. Left: 2D density plots showing a comparison of the total oxygen abundances obtained in this work from the direct Te-based method (x-axis) and those obtained from different strong-line methods (y-axis), as indicated in each panel (from left to right and from top to bottom: HCM, N2, O3N2-PP04, O3N2-M13, N2O2, R23-MG91 and R-PG16). The dotted–dashed line marks the 1:1 relation. The Spearman’s rank correlation coefficient, ρ, is shown in all the panels. The number of H II region data used for each panel is indicated in the top-left corner. Right: Difference between the O/H abundances obtained from strong lines and the corresponding Te-based value (residuals) as a function of the Te-based O/H abundances. The median value and the standard deviation (rms) for the residuals are shown in the lower left corner of each panel, together with the Spearman’s rank correlation coefficient. this calibration only in the range of validity specified by the authors not unexpected and has been previously reported (Perez-Montero´ (Marino et al. 2013). For the N2 calibration by PP04, and the R & Contini 2009), as N2O2 is also a good tracer of N/O. However, calibration by PG16 the residuals also show a moderate correlation the observed dependence does not produce an overestimation of (ρ =−0.59 and −0.56, respectively) with the Te-based oxygen 12+log (O/H) for regions of high N/O, but an underestimation abundance. For N2O2, R23, and O3N2 (as calibrated by PP04), only for regions with log (N/O) below ∼−0.9, with respect to the ones a weak dependence with the oxygen abundance is observed. obtained with the Te-based method. However, the slope in this plot The left-hand panel of Fig. 6 shows the same residuals in the is not very high and in practice, this would imply an underestimation oxygen abundances for all the strong-line methods but now as a of up to ∼0.2–0.3 dex for the lowest N/O H II regions [log (N/O)  function of the Te-based N/O abundance. Here no method shows −1.5] that is close to the typical uncertainties in strong-line methods, a significant dependence with N/O, except N2O2 for which we with standard deviations of ∼0.25 dex. Surprisingly, no dependence estimate a Spearman’s coefficient of ρ = 0.58. This behaviour is is observed in the residuals of the N2 method with log (N/O). Finally,

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2369 Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 6. Same as Fig. 5 but for the residuals against the Te-based log (N/O) abundances derived in this work (left) and log O32 (right) for the strong-line methods indicated in the top-right corner in each panel. the residuals in 12+log (O/H) are plotted as a function of O32 in the all the different strong-line methods mentioned in Section 4.3, in right-hand panel of Fig. 6. Very weak dependences are observed for order to allow for comparisons with previous results and to evaluate all methods except for O3N2 as calibrated by PP04 ρ =−0.52, a the influence of the methodology on the O/H abundance we derive. dependence already reported by Ho et al. (2015). The top panels of Fig. 7 show a comparison of the log (N/O) In summary, overall, all the strong-line methods tested here obtained with HCM, N2O2, N2S2, and the R calibration, with the globally reproduce Te-based oxygen abundances in our sample of H II values obtained from the Te-based method for our sample of H II regions (with the exception of R23), but the residuals in 12+log (O/H) regions. The corresponding residuals are plotted in the mid- and correlate to some degree with log (N/O) and/or log (O32). These cor- lower panels as a function of log (N/O) and log (O32), respectively. relations imply over- or underestimations in 12+log (O/H) depending The dispersion of the residuals is lower for the R calibration, on the method. The correlation with O32 underlines the importance HCM, and N2O2 (∼0.09–0.13 dex, cf. 0.21 dex for N2S2), but of the ionization parameter and its effects on strong-line chemical the N/O abundance ratio obtained from N2O2 is systematically abundances. The residuals are within typical uncertainties inherent to larger than the Te-based N/O. In fact, the residuals of N2O2 show = the methods, except for O3N2 as calibrated by M13, but can produce a positive correlation with log (N/O)Te , with ρ 0.55, implying systematic effects in radial abundance gradients. As these methods average overestimations in log (N/O) as large as ∼0.5 dex for the are widely used, we will compute the radial metallicity profiles for H II regions with the largest N/O abundance ratio [log (N/O) ∼

MNRAS 500, 2359–2379 (2021) 2370 A. Zurita et al. Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 7. Top: Comparison of the Te-based N/O abundances derived in this paper with those obtained from HCM, N2O2, N2S2, and the R calibration. Centre: Differences between N/O strong-line and Te-based N/O abundances (from left to right: HCM, N2O2, N2S2, R) as a function of the Te-based N/O abundance. Bottom: Same residuals in N/O abundance as a function of log (O32) for the four methods. The median value and standard deviation (rms) for the residuals are shown in the lower left corner of the central and bottom panels, together with the Spearman’s rank correlation coefficient.

−0.5]. The N2O2 method also yields residuals in log (N/O) that regions. The figures contain different panels that show the log (N/O) anticorrelate with log (O32). The residuals obtained with HCM, radial profile (from HCM) and the metallicity profiles obtained for N2S2, and the R calibration do not show important correlations five different strong-line methods: HCM, O3N2 (PP04), N2 (PP04), with log (N/O) or log (O32), with |ρ| < 0.4, but the R calibration N2O2 (B07), and R23 (MG91, as parametrized by Kobulnicky et al. and HCM are those that better match the Te-based method, with low 1999), as described in Section 4. Different colours and symbols have median residuals (<0.1 dex versus 0.23 dex for N2S2) as well as the been employed for data from different authors as shown in the legend low dispersion around the 1:1 relation already mentioned. Therefore, (upper left corner of the galaxy image). H II regions located within the N/O abundance ratios obtained with HCM and the R calibration galactic bars have been marked with a dark grey edge. Black small are those that better follow the Te scale for our H II region sample. open symbols represent Te-based abundance estimates. Vertical lines In what follows, we will then focus on the log (N/O) estimates from in the radial profiles mark the location of re (grey dashed line), r25 HCM and the R calibration, although we will derive radial profile fits (grey dotted line), and the deprojected bar radius (dark blue dashed– for all the methods. Previous works on radial abundance gradients dotted line), for barred galaxies. Given the inhomogeneous nature of in barred galaxies have been mostly concentrated on O/H; therefore, the data, we do not have the same emission lines for all regions, and there is not much work to compare with, with the exception of Perez-´ therefore in some cases we could not obtain the profiles with all the Montero et al. (2016), that uses HCM. methods due to the lack of the relevant line fluxes. A first look at the profiles shows, as expected, a wide range of abundance profile slopes (e.g. Moustakas et al. 2010; Pilyugin et al. 5 RADIAL ABUNDANCE PROFILES 2014) and radial breaks in the outer parts of some of them (e.g. In order to derive the O/H and N/O radial abundance profiles for Goddard et al. 2011; Zahid & Bresolin 2011; Bresolin et al. 2012; our sample of galaxies, we have used our recalculated deprojected Croxall et al. 2016). In about 75 per cent of the galaxies, our derived galactocentric distances (Section 3.2) and our derived oxygen and profiles cover the galactic disc beyond approximately r25, while in six nitrogen abundances for all the H II regions in the compilation, as galaxies (12 per cent of the sample) the collected data cover only up ∼ described in Sections 4.2 and 4.3. to 0.6–0.8 times r25. In order to characterize the abundance radial + The profiles are shown in Figs E1–E51 in Appendix E (available profiles, we performed a least-squares linear fit to 12 log (O/H) and online) for all galaxies and for a few selected methods, where we log (N/O) as a function of galactocentric radius (R), as parametrized + = + = + also show an image for each galaxy and the location of the H II by 12 log (O/H) bO/H αO/H R and log (N/O) bN/O αN/O R,

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2371 respectively. We performed an unweighted fit to the data, as flux line methods used in this work. In our sample, if a break is detected, error estimates are highly heterogeneous among the different authors it will probably be seen at least in the profiles derived from N2O2 and and were not always provided. All available H II regions in a galaxy R23. NGC 3031 appears to show a break (in R23 and possibly in the were used for the linear fitting, except those described in Section 3.3 N2O2 profiles) at a radius of ∼0.7 × r25, but to our knowledge, this and regions with peculiar metallicities already identified in previous has not been previously reported. In general, the break radius varies studies, as is the case of regions located in the bridge structure in between 0.5 (for NGC 3359) and 1.4 × r25 (for NGC 5236) with an NGC 1512 (Bresolin et al. 2012). The regions excluded from the average of (0.9 ± 0.3) × r25. The break radius changes among calibra- fit are marked in Figs E1–E51 with a blue edge. In two galaxies tions, but the maximum difference is ∼25 per cent in a given galaxy.

(NGC 2403 and NGC 2903), we excluded data from Smith (1975) The profile break was first detected in the 12+log (O/H) radial Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 that systematically showed a larger deviation from the bulk of the profiles of spirals, but a break in the log (N/O) radial profile was data. A minimum of five data points per profile were required for the also detected in a few galaxies (Bresolin et al. 2009a;Bergetal. fitting (Zaritsky et al. 1994). The slopes of the profiles normalized to 2013;Lopez-S´ anchez´ et al. 2015; Croxall et al. 2016;Bergetal. re and r25 were also calculated. 2020). In the case of NGC 628 and NGC 5457 (M101), a break When the profiles showed evidence for a radial abundance gradient was detected in the radial log (N/O) profile that was not seen in change or break after visual inspection, and/or the break has been 12+log (O/H). All the galaxies in this sample presenting radial breaks reported by other authors (Bresolin et al. 2009a, 2012; Goddard et al. in the 12+log (O/H) profile, with one or several methods, also show a 2011; Zahid & Bresolin 2011;Bergetal.2013; Croxall et al. 2016; break in the log (N/O) profile with either HCM, or the R calibration or Bresolin 2019), we have performed both a single linear fit and a with both of them, at approximately the same galactocentric radius. double linear fit to the radial profile. The single linear fit to the data It is also interesting to note that for the four galaxies in which a radial points is shown with a blue solid straight line in Figs E1–E51. The break is detected in the log (N/O) profile derived from the Te-based 2 corresponding slope and reduced chi-square,χν , values are shown, abundances (NGC 628, NGC 3621, NGC 5194, and NGC 5457), in the same colour, in the upper side for each panel. Double linear such a break is not detected in the Te-based 12+log (O/H) profile. 2 fits are shown with a dotted red line. The associated χν and the Given the lower dispersion in the log (N/O) profiles, the break seems improvement with respect to the single linear fit (in parenthesis) are to be more prominent and easily detected than that in 12+log (O/H). also shown in red. When this improvement is at least 10 per cent, However, a deeper analysis would be necessary. 2 the inner slope value is also given. This criterion in χν is somewhat arbitrary, but it reflects numerically the visual detection of breaks. In 2 5.2 Comparison between Te- and strong-line-based slopes most cases, the improvement in χν is very significant (>25 per cent). When T -based abundances could be derived for a number of e In Fig. 8, we compare Te-based radial abundance gradient slopes regions in a galaxy, we also derived the best linear fitting, which with those obtained from strong-line abundances for O/H (first is represented with a black dotted straight line in the profiles, seven panels, with red circles) and N/O (last two panels, with green with the slope and corresponding uncertainty also shown in black. squares). We have restricted the comparison to the better determined Unfortunately, radial abundance profiles based on the T -based e Te-based slopes, and therefore we consider only galaxies with at least abundances could be reliably derived for only 13 galaxies. Although 10 H II regions having Te-based abundances well distributed across we cannot base our study on these galaxies alone, this subsample their discs, 13 galaxies in total.12 As the radial breaks are not always offers the possibility of testing metallicity gradients derived from detected with all methods, we are comparing the slopes resulting strong-line indicators (Section 5.2). For four of the galaxies with from the single linear fit to all the radial abundance profiles. Te-based abundance profiles (NGC 628, NGC 3621, NGC 5154, and Each panel in Fig. 8 shows the metallicity gradient as derived NGC 5457), we could attempt a double linear fit, which results in a with a strong-line method, as a function of the slope derived with significant (χ 2 improvement >10 per cent) break in the log (N/O) −1 ν the Te-based method (in dex kpc ). The solid 1:1 line shows where radial profile. The fit is shown with dashed green lines and the the points would lie if the strong-line- and Te-based methods gave corresponding inner slope is shown with the same colour in the the same slope. The dotted lines mark the rms deviations from corresponding plots. the 1:1 relation indicated in the lower right corner in each panel. Electronic tables with the radial abundance fit coefficients for all The lowest rms values for the O/H slopes, αO/H, are found for the the methods, together with the corresponding correlation parameter, R calibration, N2O2 and N2 (0.009–0.013 dex kpc−1), while the 2 −1 χν , and scatter in the fits, are made publicly available on the CDS highest rms is found for O3N2 (0.017 dex kpc ). We have labelled VizieR facility for all galaxies. in each panel the galaxies with deviations from equality larger than the rms, considering error bars. NGC 5194 has a large deviation 5.1 Radial gas abundance breaks with HCM, N2, and the R calibration. In fact, all galaxies exhibit a negative oxygen abundance gradient with all methods, with the The presence of a radial break or a change in the slope of the radial exception of this galaxy, for which a positive gradient is obtained metallicity profile is a well-known feature of galaxies having an with N2. This might be produced by the difficulty of strong-line extended disc of star formation (see Bresolin 2017,forareview).For methods in giving good estimates of metallicity for low-excitation seven galaxies of our sample (NGC 1058, NGC 1512, NGC 3359, H II regions (Bresolin et al. 2004). This is evident in Fig. E39 NGC 3621, NGC 4625, NGC 5236, and NGC 5457), radial breaks (available online) where we can see that most of the strong-line in the 12+log (O/H) profile, at a galactocentric position close to methods yield low abundances for the innermost regions. The latter the isophotal radius r25, have been reported by previous work (see Table A1 for the references). We confirm with our re-analysis of the compiled data the presence of such breaks in 12+log (O/H) via an im- 12 The Te-based slopes for NGC 1512, NGC 2997, and NGC 4258 were not 2 provement in χν in the double linear fit with respect to the single lin- taken into account for this comparison because they were determined by a ear fit to the radial profile (see Section 5 and Figs E1–E51, available smaller number of data points and/or the data points are not well distributed online). However, the break is not clearly detected with all the strong- across the disc and show a large dispersion.

MNRAS 500, 2359–2379 (2021) 2372 A. Zurita et al. Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

Figure 8. Comparison of slopes obtained from different strong-line methods as a function of those derived from Te-based abundances for the subsample of galaxies with reliable Te-based abundance profiles. The first seven panels show the comparison for slopes of the oxygen abundance radial profile (αO/H,red circles). The slopes of the log (N/O) radial profile (αN/O, green squares) are shown in the two bottom right panels. The solid grey line marks the 1:1 relation. The dotted lines show the rms dispersion about equality. The rms is also shown in the lower right corner of each panel together with the Spearman’s correlation coefficient (ρ). The galaxies with larger deviations from the 1:1 line are labelled. are excluded from the fits (see Section 3.3), and this does not alter the may produce a smaller variation in O3N2 oxygen abundances across metallicity gradient we derive except for the profiles obtained with the galactic disc in the galaxy, which would imply a smaller derived N2 and HCM, for which we found a systematic trend of decreasing radial abundance gradient. The importance of this method-induced oxygen abundances with decreasing distance to the galaxy centre flattening would of course depend on the average metallicity of the from galactocentric distances as large as ∼4 kpc. The N2, HCM, and galaxy, but it would be expected to affect more those galaxies with R23 methods are not able to estimate well the oxygen abundance of larger gradients or with a larger range of real oxygen abundances the H II regions with low ionization degree in the Milky Way either across their discs. This is exactly what we see in the third and fourth (Esteban & Garc´ıa-Rojas 2018), but in this particular case the slope panels in Fig. 8, especially for the M13 calibration. Shallower remains negative, although the gradient is considerably shallower slopes in radial abundance gradients derived using O3N2 were also with N2 and R23. found by Bresolin & Kennicutt (2015) in a sample of low-surface Fig. 8 also shows a tendency for methods based on the O3N2 brightness galaxies (LSBGs), and also by Erroz-Ferrer et al. (2019) indicator to yield shallower oxygen abundance gradients as for a sample of 38 spirals observed with MUSE. compared with the direct method. This is probably explained by Overall, the metallicity gradients derived with HCM, N2, N2O2, the fact already discussed in Section 4.4 that oxygen abundances and the R calibration show a moderate-to-strong correlation with obtained from O3N2, notably with the M13 calibration, are restricted the slopes derived with the Te-based method. The Spearman’s rank to a limited range in 12+log (O/H), smaller than the corresponding correlation coefficients for these methods are larger than ∼0.5, with values obtained from the Te-based method for the same regions. This the strongest correlation for R andHCM,withρ = 0.73 and 0.62, effect, translated to the H II region abundances in a given galaxy, respectively, for this sample of galaxies. The slopes derived with the

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2373

Table 2. Published values of the mean (or median) slopes of the 12+log (O/H) and log (N/O) radial abundance profiles from different data samples and methodologies as indicated.

a αO/H Indicator and calibration Sample and reference − ± −1 = 0.11 0.09 dex re O3N2 (PP04) N 193 from the CALIFA sample, Sanchez´ et al. (2014) − ± −1 = 0.075 0.016 dex re O3N2 (M13) N 122 from the CALIFA sample, Sanchez-Menguiano´ et al. (2016) − ± −1 = 0.10 0.03 dex re O3N2 (M13) N 102 from the AMUSING survey, Sanchez-Menguiano´ et al. (2018) − ± −1 = 0.076 0.064 dex re O3N2 (PP04) N 10, low-surface brightness galaxies, Bresolin & Kennicutt (2015) −1

− ± = Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 0.13 0.06 dex re C method (P14) N 46 from the Pilyugin et al. (2014) sample, Bresolin (2019) − ± −1 = 0.053 0.068 dex re HCM (PM14) N 201 from the CALIFA sample, Perez-Montero´ et al. (2016) − ± −1 = 0.13 0.07 dex re R calibration (PG16) N 147 from the MaNGA survey, calculated from data in Pilyugin et al. (2019)

− ± −1 = 0.39 0.18 dex r25 N2O2 (KD02) N 49, local field star-forming galaxies, Ho et al. (2015) − ± −1 = 0.133 0.072 dex r25 N2O2 (B07) N 10, low-surface brightness galaxies, Bresolin & Kennicutt (2015) − ± −1 = 0.32 0.20 dex r25 C method (P14) N 104 from the Pilyugin et al. (2014) sample, Ho et al. (2015) − ± −1 = 0.21 0.09 dex r25 R calibration (PG16) N 147 from the MaNGA survey, calculated from data in Pilyugin et al. (2019)

a αN/O Indicator and calibration Sample and reference − ± −1 = 0.104 0.096 dex re HCM (PM14) N 201 from the CALIFA sample, Perez-Montero´ et al. (2016) − ± −1 = 0.33 0.08 dex re Te-based N 4 (NGC 628, NGC 5194, NGC 5457, NGC 3184), Berg et al. (2020) Note. aPP04: Pettini & Pagel (2004), M13: Marino et al. (2013), P14: Pilyugin et al. (2014), KD02: Kewley & Dopita (2002), B07: Bresolin (2007), PM14: Perez-Montero´ (2014), PG16: Pilyugin & Grebel (2016).

O3N2 indicator as calibrated by M13 are the ones that more weakly Zaritsky et al. 1994). The result comes from the analysis of IFU correlate with the Te-based ones (ρ = 0.13). We are aware that the spectra of different data samples: CALIFA in Sanchez´ et al. (2014) above comparison must be interpreted with caution. First, it is based and Sanchez-Menguiano´ et al. (2016) and MUSE-VLT in Sanchez-´ on a small data sample. Secondly, the number of data measurements Menguiano et al. (2018), with the same metallicity indicator, O3N2, that define the Te-based slope is typically lower (by even a factor of and normalization of the gradients (disc effective radius). The latter − ± −1 10) than the number of measurements defining the strong-line method work found 0.10 0.03 dex re . Ho et al. (2015) analysed 49 slope. This can give way to biases. In spite of this, it is worth doing the local field galaxies with the O3N2 and N2O2 calibrations, and also comparison. The number of works in which strong-line methods are report the existence of a common slope when metallicity gradients − ± −1 used to obtain radial abundance profiles in large samples of galaxies are normalized by the disc isophotal radius: 0.39 0.18 dex r25 . is rapidly increasing. Currently, this is probably one of the most Table 2 shows median or average slopes obtained by different authors complete data sets for comparing different metallicity estimates, and for different galaxy samples and metallicity scales. scrutinizing strong-line methods against the direct method provides The αO/H and αN/O median values from our galaxy sample are a useful and healthy check on their reliability until better data sets shown in Table 3 for the different diagnostics employed in this paper, −1 −1 −1 become available. in units of dex kpc ,dexr25 , and dex re . The number of galaxies The two bottom-right panels of Fig. 8 show that the HCM and used in the statistics varies between 42 and 51 depending on the R calibration slopes for the log (N/O) radial profiles are in good strong-line method. We also show the median value derived for the agreement with the Te-based gradients. The dispersion is similar to 13 galaxies for which the Te-based abundance profile is reliable the one obtained from the 12+log (O/H), 0.013 and 0.009 dex kpc−1, (see Section 5.2). The slope values are graphically shown in Figs 9 and the Spearman coefficients (ρ = 0.79 and 0.91) indicate a strong and 10 for 12+log (O/H) and log (N/O), respectively. These contain correlation between the two slopes for the two methods. violin plots with the distribution of slope values derived for the Therefore, based on both the comparison of the strong-line different methods (individual grey dots inside the violins), together abundances with the Te-based ones for individual H II regions with the probability density of the data at different slope values (the (Section 4.4), and the comparison of the radial abundance gradients, coloured violins themselves). The probability density is a kernel the R calibration, and the HCM methods are those that better match density estimate of the underlying distribution.13 Median values for Te-based 12+log (O/H) and log (N/O) abundances in our sample. the corresponding distributions are marked with a white star for each However, as we ultimately aim to understand the causes behind the metallicity method. The two panels in Fig. 9 correspond to metallicity contradicting results on the effect of bars on the metallicity gradients, slopes normalized to r25 and re in the top and bottom panels, and these have been derived from a variety of strong-line methods, respectively. For comparison purposes, we also show the values we will continue using and analysing O/H and N/O results obtained reported by other authors from different samples and metallicity from several methods. These are presented in the following sections. calibrations (Table 2). Both in Table 3 and Figs 9 and 10,we have considered the inner slope in the profiles with radial breaks, as explained in Section 5. 5.3 A characteristic O/H and N/O abundance gradient? Fig. 9 nicely shows that the distribution of metallicity gradients is clearly dependent on the method in both the median value and Recent work on the gas-phase radial metallicity gradients in spirals the standard deviation. For all the strong-line methods, we obtain a has considerably enlarged the galaxy sample sizes, yielding impor- standard deviation in the slope distribution that ranges from ∼0.09 tant results, such as the confirmation of an earlier proposal for the existence of a universal metallicity gradient in spirals when slopes are normalized to galaxy size (e.g. Vila-Costas & Edmunds 1992; 13We employed the seaborn PYTHON module.

MNRAS 500, 2359–2379 (2021) 2374 A. Zurita et al.

Table 3. Median slope of the radial oxygen abundance gradient in different diversity of gradients found with just four galaxies from the CHAOS units for the galaxies analysed in this paper. The uncertainties show the project for which the slopes have been derived very reliably with standard deviation in the slope values. the Te-based method. Our Te-based slopes support their statement. However, Berg et al. (2020) find that the same four galaxies have log (O/H) versus R a very similar value of αN/O in terms of re, for gradients measured a Method αO/H αO/H αO/H N between 0.2 and 2 times re. We show in Fig. 10 the distribution of −1 −1 −1 α values for our sample galaxies from the T -based method and (dex kpc )(dexr25 )(dexre ) N/O e from both the HCM and the R calibration. The rms dispersion of the HCM (PM14) − 0.01 ± 0.03 − 0.17 ± 0.33 − 0.05 ± 0.12 51 α value distributions is comparable to those for α (σ = 0.12– Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 − ± − ± − ± N/O O/H R23 (M91) 0.03 0.03 0.39 0.26 0.15 0.10 48 0.15 dex r−1). Taking into account uncertainties, our median value for N2O2 (B07) − 0.03 ± 0.03 − 0.44 ± 0.26 − 0.15 ± 0.11 43 e HCM also agrees with the one derived by Perez-Montero´ et al. (2016) N2 (PP04) − 0.01 ± 0.02 − 0.16 ± 0.22 − 0.06 ± 0.09 44 O3N2 (PP04) − 0.03 ± 0.03 − 0.39 ± 0.27 − 0.16 ± 0.11 44 with the same method but for a different and much larger sample of O3N2 (M13) − 0.02 ± 0.02 − 0.24 ± 0.18 − 0.10 ± 0.07 44 201 galaxies from CALIFA, but Perez-Montero´ et al. (2016) obtained −1 R (PG16) − 0.02 ± 0.02 − 0.27 ± 0.23 − 0.10 ± 0.09 42 a smaller dispersion of 0.096 dex re . Our median value derived from b Te-based − 0.03 ± 0.02 − 0.41 ± 0.25 − 0.17 ± 0.08 13 the 13 galaxies with reliable Te-based radial log (N/O) abundance profiles is, however, considerably shallower than the characteristic − ± −1 log (N/O) versus R value of 0.33 0.08 dex re derivedbyBergetal.(2020). However, 14 a our median αN/O value for the four CHAOS galaxies included in Method αN/O αN/O αN/O N −1 − Berg et al. (2020), −0.41 ± 0.08 dex r (violet star in Fig. 10), is (dex kpc−1)(dexr 1)(dexr−1) e 25 e closer to their reported characteristic value. Part of the discrepancy HCM (PM14) − 0.04 ± 0.04 − 0.5 ± 0.3 − 0.18 ± 0.15 45 between our and their median value for these four galaxies might R (PG16) − 0.04 ± 0.02 − 0.4 ± 0.2 − 0.15 ± 0.12 42 come from differences in the inner slope for NGC 5194, for which T -basedb − 0.05 ± 0.02 − 0.5 ± 0.3 − 0.18 ± 0.15 13 − ± −1 e we derive a steeper slope of 0.53 0.16 dex re than the cited − a − ± 1 Notes. Acronyms as follows: PM14: Perez-Montero´ (2014), M91: McGaugh authors do ( 0.17 0.11 dex re ), possibly due to different adopted (1991), B07: Bresolin (2007), values of re for this galaxy. PP04: Pettini & Pagel (2004), M13: Marino et al. (2013), and PG16: Pilyugin Our galaxy sample was not designed to test the existence of a uni- &Grebel(2016). versal abundance gradient in spirals and might not be appropriate for b Median value for the galaxies with more reliable determination of the this aim. However, the reasonable agreement of our data with values gradients (i.e. NGC 1512, NGC 2997, and NGC 4258 were excluded). derived from previous studies, and the large measured dispersion of the slope distributions, might indicate that if a universal gradient −1 ∼ −1 to 0.12 dex re or from 0.22 to 0.33 dex r25 , except for O3N2 as exists, second-order effects might be important in the modification of calibrated by Marino et al. (2013), for which it is systematically lower the characteristic gradient, and they might be relevant in producing −1 −1 with the two normalizations (i.e. 0.07 dex r25 and 0.17 dex re ). This the observed dispersion of slope values for both αO/H and αN/O. is not unexpected, given our analysis in Sections 4.4 and 5.2, where In Paper II, we will analyse the dependence of the radial abundance our comparison of direct and strong-line methods showed a tendency gradients of O/H and N/O with host galaxy properties, in particular for this calibration of the O3N2 index to flatten the steepest gradients, with the presence of a galactic stellar bar. due to the small range in the O3N2-predicted metallicities in relation to the Te-based ones (Fig. 5). This implies that gradients derived with this method tend to cover a smaller range of values. Therefore, the 6 SUMMARY AND CONCLUSIONS O3N2 method, notably with the calibration performed by Marino This paper is the first of a series of two, devoted to revisit the issue et al. (2013), favours the conclusion of the existence of a universal of the effect of stellar bars on the radial distribution of metals in metallicity gradient, but the dispersion in slope values may be biased the gas-phase of spiral galaxies, as an attempt to solve the existing by the method selection. discrepancies found in the literature. For that aim, we have made a The HCM and N2 methods yield a shallower median slope than the compilation of published emission-line fluxes and positions of over rest of methods by an amount similar to the standard deviation of the 2800 H II regions, which belong to a sample of 51 nearby (D < ∼ −1 ∼ −1 distributions ( 0.1 dex re or 0.2 dex r25 ). Regarding the median 64 Mpc) spiral galaxies. We have used a homogeneous methodology slope values, it is interesting to note that there is good agreement to derive: (a) the structural parameters for bars and discs from broad- between the median slopes derived by us with a given method and band imaging, and (b) O/H and N/O abundance ratios from the direct the corresponding value obtained for a different sample by other or Te-based method for a subsample of regions (610), and from authors with the same method. The metallicity method employed a variety of strong-line methods [based on the R calibration, R23, seems to have a stronger effect on the differences in median values N2O2, O3N2, N2, and HCM for 12+log (O/H), and for log (N/O) between works than differences in the galaxy sample and/or in the on N2O2, N2S2, the R calibration and HCM]. The abundances nature of the observational data. The largest difference is seen in the derived from the strong-line methods have been compared to the −1 median slope in dex r25 derived from the N2O2 scale between the more reliable Te-based ones for the subsample of regions for which sample of LSBGs in Bresolin & Kennicutt (2015), violet circle in the direct abundances could be calculated. Radial 12+log (O/H) and top panel of Fig. 9, and our sample or the one by Ho et al. (2015), log (N/O) abundance profiles have been derived for all galaxies from both composed by high-surface brightness galaxies. This difference re-calculated deprojected H II region galactocentric distances for all between the slopes in low- and high-surface brightness galaxies, not the abundance methods and have been analysed. Our main results −1 clearly seen when slopes are normalized to re with the O3N2 scale, and conclusions are summarized below: is deeply discussed in Bresolin & Kennicutt (2015). Recent work by Berg et al. (2020) questions the existence of a universal gradient in 12+log (O/H). This conclusion comes from the 14NGC 628, NGC 3184, NGC 5194, and NGC 5457.

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2375 Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

−1 −1 Figure 9. Violin plots showing the distribution of the metallicity slopes derived with the different methods in units of dex r25 (top) and dex re (bottom). The grey points inside the violin plots represent the slope values for the individual galaxies of the sample, and the white star represents the median slope for each method. For galaxies presenting metallicity breaks in their profiles, we have considered the inner slope value. Average or median values obtainedby other authors are shown in the right-hand side of each panel for comparison. These include Ho et al. (2015,Ho+15), Pilyugin et al. (2014,P+14), Bresolin & Kennicutt (2015, BK15), Sanchez´ et al. (2014,SF+14), Sanchez-Menguiano´ et al. (2016,SM+16), Sanchez-Menguiano´ et al. (2018,SM+18), Perez-Montero´ et al. (2016,PM+16), Bresolin (2019, B19), and Pilyugin et al. (2019,P+19).

(i) The RC3 visual classification of bars should be taken with (iv) Based on both the comparison of Te-based with strong-line caution. In our sample, the ‘AB’ class contains galaxies with a metallicities for individual H II regions, and the comparison of wide range in bar ellipticity (a proxy for bar strength). Using this the metallicity gradients derived from both methodologies, the R parameter, our sample contains 22 strongly barred, 9 weakly barred calibration, HCM, and N2O2 yield a better agreement with the Te- and 20 unbarred galaxies. based scale for our H II region sample. (ii) Overall, all strong-line methods tested here reproduce the Te- (v) For log (N/O), the HCM method and the R calibration are those based abundances [except for the well-known offset for the R23 that better match the Te-based scale for individual H II regions. These method calibrated from photoionization models from McGaugh also reproduce well the Te-based radial log (N/O) slopes. (1991)]. However, we find that the residuals correlate to some extent (vi) We confirm the presence of breaks in the radial abundance + with 12 log (O/H)Te , log (N/O)Te , and/or log (O32) for some of the profiles of the galaxies analysed, in agreement with previous work methods. Special caution must be taken with the O3N2 method, by different authors. However, the breaks in the 12+log (O/H) are notably with the calibration performed by Marino et al. (2013), that not clearly detected with all the strong-line methods. In our sample, yields a smaller range in metallicities than the Te-based one. if a break is detected, the N2O2 and R23 methods are more prone to (iii) The above result translates in a tendency for this method to show it. flatten the radial metallicity profiles of the galaxies with the steepest (vii) Most of the galaxies with a break in their metallicity profile gradients in our sample. This is clearly seen in our comparison of (with one or several methods) also show a break in their log (N/O) slopes derived from strong-line methods with those obtained from radial profile at a similar radius. The break radius [in either the the Te-based method for a subsample of 13 galaxies with reliable 12+log (O/H) or the log (N/O) profiles] is typically in the range determination of the Te-based metallicity gradient. 0.5–1.4 times r25 with an average of (0.9 ± 0.3) × r25.

MNRAS 500, 2359–2379 (2021) 2376 A. Zurita et al.

Astropy Collaboration 2018); Matplotlib (Hunter 2007); APLpy, an open-source plotting package for PYTHON (Robitaille & Bressert 2012);NumPy(vanderWaltetal. 2011) and seaborn (DOI: 10.5281/zenodo.1313201). We acknowledge the usage of the Hyper- Leda database (http://leda.univ-lyon1.fr). This research has made use of the VizieR catalogue access tool, CDS, Strasbourg, France (DOI: 10.26093/cds/vizier). The original description of the VizieR service was published in 2000, A&AS 143, 23. This research has

made use of the NASA/IPAC Extragalactic Database (NED) that is Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

DATA AVAILABILITY The data underlying this article are available in the article, in its online supplementary material, and on the CDS VizieR facility (via Figure 10. Violin plot showing the distribution of the log (N/O) abundance https://vizier.u-strasbg.fr/viz-bin/VizieR). gradient derived with HCM, the R calibration, and the Te-based method in −1 units of dex re . For comparison, we also show the median slope value derived from the Te-based method in this work for the four CHAOS galaxies REFERENCES included in Berg et al. (2020), violet star, and the median values derived by Perez-Montero´ et al. (2016) and Berg et al. (2020). Abraham R. G., Merrifield M. R., 2000, AJ, 120, 2835 Abraham R. G., Merrifield M. R., Ellis R. S., Tanvir N. R., Brinchmann J., (viii) We have analysed the median slope and dispersion values 1999, MNRAS, 308, 569 derived for 12+log (O/H) and log (N/O) for the different methods. Aguerri J. A. L., Munoz-Tu˜ n˜on´ C., Varela A. M., Prieto M., 2000, A&A, 361, Both the median and the dispersion depend on the selected method 841 to derive the abundances, but are in reasonable agreement with Aller L. H., 1942, ApJ, 95, 52 results derived by other authors from different data samples. The Alloin D., Edmunds M. G., Lindblad P. O., Pagel B. E. J., 1981, A&A, 101, O3N2 method as calibrated by Marino et al. (2013) yields a smaller 377 Arellano-Cordova´ K. Z., Rodr´ıguez M., Mayya Y. D., Rosa-Gonzalez´ D., dispersion than the rest of methods with the two normalizations of the 2016, MNRAS, 455, 2627 profiles (with re and r25), which supports the proposal of a universal Astropy Collaboration, 2013, A&A, 558, A33 metallicity gradient. However, the low measured dispersion in the Astropy Collaboration, 2018, AJ, 156, 123 slope distribution might be biased by the method selection and the Athanassoula E., 2003, MNRAS, 341, 1179 associated problems with O3N2 metallicity scale mentioned above Baillard A. et al., 2011, A&A, 532, A74 (in comparison with the Te-based abundances). Baldwin J. A., Phillips M. M., Terlevich R., 1981, PASP, 93, 5 Beaton R. L. et al., 2007, ApJ, 658, L91 In Paper II, we analyse the derived radial abundance gradients Berg D. A., Pogge R. W., Skillman E. D., Croxall K. V., Moustakas J., Rogers comparatively for (strongly and weakly) barred and unbarred galax- N. S. J., Sun J., 2020, ApJ, 893, 96 ies. The results are compared with previous work, and discussed in Berg D. A., Skillman E. D., Croxall K. V., Pogge R. W., Moustakas J., the context of current knowledge on disc evolution and radial mixing Johnson-Groh M., 2015, ApJ, 806, 16 induced by bars and spiral arms. Berg D. A., Skillman E. D., Garnett D. R., Croxall K. V., Marble A. R., Smith J. D., Gordon K., Kennicutt Robert C. J., 2013, ApJ, 775, 128 Bibby J. L., Crowther P. A., 2010, MNRAS, 405, 2737 ACKNOWLEDGEMENTS Blair W. P., Kirshner R. P., Chevalier R. A., 1982, ApJ, 254, 50 Bland-Hawthorn J., Gerhard O., 2016, ARA&A, 54, 529 We kindly thank all the authors who have previously published Blana˜ D´ıaz M., Gerhard O., Wegg C., Portail M., Opitsch M., Saglia R., the H II data that form the basis for this work, and Simon Verley Fabricius M., Erwin P., Bender R., 2018, MNRAS, 481, 3210 for his invaluable help with PYTHON. We also acknowledge the Blana˜ D´ıaz M., Wegg C., Gerhard O., Erwin P., Portail M., Opitsch M., anonymous referee for his/her suggestions that improved the clarity Saglia R., Bender R., 2017, MNRAS, 466, 4279 of the manuscript. We thank Calar Alto Observatory for the allocation Bovy J., Rix H.-W., 2013, ApJ, 779, 115 of director’s discretionary time to this programme. AZ, EF, and IP Bresolin F., 2007, ApJ, 656, 186 acknowledge support from the Spanish ‘Ministerio de Econom´ıa, Bresolin F., 2011a, ApJ, 729, 56 Industria y Competitividad’ (MINECO) and from the European Bresolin F., 2011b, ApJ, 730, 129 Regional Development Fund (FEDER) via grant AYA2017-84897- Bresolin F., 2017, in Gil de Paz A., Knapen J. H., Lee J. C., eds, Proc. IAU Symp. 321, Formation and Evolution of Galaxy Outskirts. Kluwer, P, and from the Junta de Andaluc´ıa (Spain) local government Dordrecht, p. 147 through the FQM-108 project. EPM acknowledges funding from Bresolin F., 2019, MNRAS, 488, 3826 the Spanish MINECO project Estallidos 6 AYA2016-79724-C4 Bresolin F., Garnett D. R., Kennicutt Robert C. J., 2004, ApJ, 615, 228 and from the Severo Ochoa excellence programme (SEV-2017- Bresolin F., Gieren W., Kudritzki R., Pietrzynski´ G., Urbaneja M. A., Carraro 0709). This research made use of Astropy,15 a community-developed G., 2009b, ApJ, 700, 309 core PYTHON package for Astronomy (Astropy Collaboration 2013; Bresolin F., Kennicutt R. C., 2015, MNRAS, 454, 3664 Bresolin F., Kennicutt R. C., Ryan-Weber E., 2012, ApJ, 750, 122 Bresolin F., Kennicutt R. C., Jr, 2002, ApJ, 572, 838 15http://www.astropy.org Bresolin F., Kennicutt R. C. , Garnett D. R., 1999, ApJ, 510, 104

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2377

Bresolin F., Ryan-Weber E., Kennicutt R. C., Goddard Q., 2009a, ApJ, 695, Fernandez-Mart´ ´ın A., Perez-Montero´ E., V´ılchez J. M., Mampaso A., 2017, 580 A&A, 597, A84 Bresolin F., Schaerer D., Gonzalez´ Delgado R. M., Stasinska´ G., 2005, A&A, Fierro J., Torres-Peimbert S., Peimbert M., 1986, PASP, 98, 1032 441, 981 Firpo V., Bosch G., Morrell N., 2005, MNRAS, 356, 1357 Bresolin F., Stasinska´ G., V´ılchez J. M., Simon J. D., Rosolowsky E., 2010, Florido E., Zurita A., Perez´ I., Perez-Montero´ E., Coelho P. R. T., Gadotti D. MNRAS, 404, 1679 A., 2015, A&A, 584, A88 Brown M. J. I. et al., 2014, ApJS, 212, 18 Frei Z., Guhathakurta P., Gunn J. E., Tyson J. A., 1996, AJ, 111, 174 Buta R. J., Corwin H. G., Odewahn S. C., 2007, The de Vaucouleurs Atlas of Gadotti D. A., 2009, MNRAS, 393, 1531 Galaxies. Cambridge University Press, Cambridge Gadotti D. A., Athanassoula E., Carrasco L., Bosma A., de Souza R. E., Cacho R., Sanchez-Bl´ azquez´ P., Gorgas J., Perez´ I., 2014, MNRAS, 442, Recillas E., 2007, MNRAS, 381, 943 Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 2496 Galarza V. C., Walterbos R. A. M., Braun R., 1999, AJ, 118, 2775 Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245 Garc´ıa-Rojas J., Esteban C., Peimbert A., Peimbert M., Rodr´ıguez M., Ruiz Castellanos M., D´ıaz A. I., Terlevich E., 2002, MNRAS, 329, 315 M. T., 2005, MNRAS, 362, 301 Cedres´ B., Cepa J., 2002, A&A, 391, 809 Garc´ıa-Rojas J., Esteban C., Peimbert A., Rodr´ıguez M., Peimbert M., Ruiz Chapelon S., Contini T., Davoust E., 1999, A&A, 345, 81 M. T., 2007, Rev. Mex. Astron. Astrofis., 43, 3 Cheng K. P., Collins N., Angione R., Talbert F., Hintzen P., Smith Garc´ıa-Rojas J., Esteban C., Peimbert M., Costado M. T., Rodr´ıguez M., E. P., Stecher T., UIT Team, eds, 1997, Uv/visible Sky Gallery Peimbert A., Ruiz M. T., 2006, MNRAS, 368, 253 on CDROM, data available from NASA Extragalactic Database Garc´ıa-Rojas J., Esteban C., Peimbert M., Rodr´ıguez M., Ruiz M. T., Peimbert (NED) A., 2004, ApJS, 153, 501 Considere` S., Coziol R., Contini T., Davoust E., 2000, A&A, 356, 89 Garc´ıa-Rojas J., Simon-D´ ´ıaz S., Esteban C., 2014, A&A, 571, A93 Cook D. O. et al., 2014, MNRAS, 445, 881 Garnett D. R., 1992, AJ, 103, 1330 Corbelli E., Lorenzoni S., Walterbos R., Braun R., Thilker D., 2010, A&A, Garnett D. R., Kennicutt Robert C. J., 1994, ApJ, 426, 123 511, A89 Garnett D. R., Shields G. A., 1987, ApJ, 317, 82 Courteau S., Widrow L. M., McDonald M., Guhathakurta P., Gilbert Garnett D. R., Shields G. A., Skillman E. D., Sagan S. P., Dufour R. J., 1997, K. M., Zhu Y., Beaton R. L., Majewski S. R., 2011, ApJ, 739, ApJ, 489, 63 20 Gil de Paz A. et al., 2007, ApJS, 173, 185 Crockett N. R., Garnett D. R., Massey P., Jacoby G., 2006, ApJ, 637, 741 Goddard Q. E., Bresolin F., Kennicutt R. C., Ryan-Weber E. V., Rosales- Croxall K. V., Pogge R. W., Berg D. A., Skillman E. D., Moustakas J., 2015, Ortega F. F., 2011, MNRAS, 412, 1246 ApJ, 808, 42 Goddard Q. E., Kennicutt R. C., Ryan-Weber E. V., 2010, MNRAS, 405, Croxall K. V., Pogge R. W., Berg D. A., Skillman E. D., Moustakas J., 2016, 2791 ApJ, 830, 4 Goetz M., Koeppen J., 1992, A&A, 262, 455 Dale D. A. et al., 2009, ApJ, 703, 517 Gonzalez Delgado R. M., Perez E., 1997, MNRAS, 284, 931 Dennefeld M., Kunth D., 1981, AJ, 86, 989 Goodwin S. P., Gribbin J., Hendry M. A., 1998, The Observatory, 118, 201 de Vaucouleurs G., de Vaucouleurs A., Corwin H. G., Jr, Buta R. J., Paturel G., Grisoni V., Spitoni E., Matteucci F., 2018, MNRAS, 481, 2570 Fouque´ P., 1991, Third Reference Catalogue of Bright Galaxies. Volume Gusev A. S., Pilyugin L. S., Sakhibov F., Dodonov S. N., Ezhkova O. V., I: Explanations and References. Volume II: Data for Galaxies Between 0h Khramtsova M. S., 2012, MNRAS, 424, 1930 and 12h. Volume III: Data for Galaxies Between 12h and 24h. Springer- Hadfield L. J., Crowther P. A., 2007, MNRAS, 381, 418 Verlag, New York Henry R. B. C., Pagel B. E. J., Chincarini G. L., 1994, MNRAS, 266, 421 Diaz A. I., Terlevich E., Vilchez J. M., Pagel B. E. J., Edmunds M. G., 1991, Henry R. B. C., Pagel B. E. J., Lasseter D. F., Chincarini G. L., 1992, MNRAS, MNRAS, 253, 245 258, 321 Dufour R. J., Talbot R. J. J., Jensen E. B., Shields G. A., 1980, ApJ, 236, 119 Ho I. T., Kudritzki R.-P., Kewley L. J., Zahid H. J., Dopita M. A., Bresolin Dutil Y., Roy J.-R., 1999, ApJ, 516, 62 F., Rupke D. S. N., 2015, MNRAS, 448, 2030 D´ıaz-Garc´ıa S., Salo H., Laurikainen E., Herrera-Endoqui M., 2016, A&A, Hoopes C. G., Walterbos R. A. M., Bothun G. D., 2001, ApJ, 559, 878 587, A160 Hunter J. D., 2007, Computing in Science & Engineering, 9, 90 D´ıaz A. I., Castellanos M., Terlevich E., Luisa Garc´ıa-Vargas M., 2000, Izotov Y.I., Stasinska´ G., Meynet G., Guseva N. G., Thuan T. X., 2006, A&A, MNRAS, 318, 462 448, 955 D´ıaz A.´ I., Terlevich E., Castellanos M., Hagele¨ G. F., 2007, MNRAS, 382, Izotov Y. I., Thuan T. X., Stasinska´ G., 2007, ApJ, 662, 15 251 James P. A. et al., 2004, A&A, 414, 23 Edmunds M. G., Pagel B. E. J., 1984, MNRAS, 211, 507 Kaplan K. F. et al., 2016, MNRAS, 462, 1642 Ellison S. L., Nair P., Patton D. R., Scudder J. M., Mendel J. T., Simard L., Kauffmann G. et al., 2003, MNRAS, 346, 1055 2011, MNRAS, 416, 2182 Kehrig C. et al., 2011, A&A, 526, A128 Erroz-Ferrer S. et al., 2015, MNRAS, 451, 1004 Kelly B. C., 2007, ApJ, 665, 1489 Erroz-Ferrer S. et al., 2019, MNRAS, 484, 5009 Kennicutt R. C., Jr, Bresolin F., Garnett D. R., 2003, ApJ, 591, 801 Erwin P., 2005, MNRAS, 364, 283 Kennicutt R. C., Jr, Keel W. C., Blaha C. A., 1989, AJ, 97, 1022 Erwin P., 2018, MNRAS, 474, 5372 Kennicutt Robert C. J., Garnett D. R., 1996, ApJ, 456, 504 Esteban C., Bresolin F., Garc´ıa-Rojas J., Toribio San Cipriano L., 2020, Kewley L. J., Dopita M. A., 2002, ApJS, 142, 35 MNRAS, 491, 2137 Kewley L. J., Dopita M. A., Sutherland R. S., Heisler C. A., Trevena J., 2001, Esteban C., Bresolin F., Peimbert M., Garc´ıa-Rojas J., Peimbert A., Mesa- ApJ, 556, 121 Delgado A., 2009, ApJ, 700, 654 Kewley L. J., Ellison S. L., 2008, ApJ, 681, 1183 Esteban C., Carigi L., Copetti M. V. F., Garc´ıa-Rojas J., Mesa-Delgado A., Knapen J. H., Stedman S., Bramich D. M., Folkes S. L., Bradley T. R., 2004, Castaneda˜ H. O., Pequignot´ D., 2013, MNRAS, 433, 382 A&A, 426, 1135 Esteban C., Fang X., Garc´ıa-Rojas J., Toribio San Cipriano L., 2017, MNRAS, Kobulnicky H. A., Kennicutt R. C., Jr, Pizagno J. L., 1999, ApJ, 514, 544 471, 987 Koribalski B. S., Lopez-S´ anchez´ A.´ R., 2009, MNRAS, 400, 1749 Esteban C., Garc´ıa-Rojas J., 2018, MNRAS, 478, 2315 Kuchinski L. E. et al., 2000, ApJS, 131, 441 Esteban C., Peimbert M., Garc´ıa-Rojas J., Ruiz M. T., Peimbert A., Rodr´ıguez Kwitter K. B., Aller L. H., 1981, MNRAS, 195, 939 M., 2004, MNRAS, 355, 229 Larsen S. S., Richtler T., 1999, A&A, 345, 59 Evans I. N., Dopita M. A., 1987, ApJ, 319, 662 Licquia T. C., Newman J. A., Bershady M. A., 2016, ApJ, 833, 220 Ferguson A. M. N., Gallagher J. S., Wyse R. F. G., 1998, AJ, 116, 673 Licquia T. C., Newman J. A., Brinchmann J., 2015, ApJ, 809, 96

MNRAS 500, 2359–2379 (2021) 2378 A. Zurita et al.

Lin Z. et al., 2017, ApJ, 842, 97 Rozas M., Zurita A., Beckman J. E., 2000, A&A, 354, 823 Lisenfeld U., Mundell C. G., Schinnerer E., Appleton P. N., Allsopp J., 2008, Rozas M., Zurita A., Heller C. H., Beckman J. E., 1999, A&AS, 135, 145 ApJ, 685, 181 Rupke D. S. N., Kewley L. J., Chien L. H., 2010, ApJ, 723, 1255 Liu X., Storey P. J., Barlow M. J., Danziger I. J., Cohen M., Bryce M., 2000, Sanders N. E., Caldwell N., McDowell J., Harding P., 2012, ApJ, 758, 133 MNRAS, 312, 585 Sanders R. L., Shapley A. E., Zhang K., Yan R., 2017, ApJ, 850, 136 Li Y., Bresolin F., Kennicutt Robert C. J., 2013, ApJ, 766, 17 Schonrich¨ R., Binney J., 2009, MNRAS, 396, 203 Luridiana V., Esteban C., Peimbert M., Peimbert A., 2002, Rev. Mex. Astron. Scoville N. Z., Matthews K., Carico D. P., Sanders D. B., 1988, ApJ, 327, Astrofis., 38, 97 L61 Lopez-Hern´ andez´ J., Terlevich E., Terlevich R., Rosa-Gonzalez´ D., D´ıaz Searle L., 1971, ApJ, 168, 327 A.,´ Garc´ıa-Benito R., V´ılchez J., Hagele¨ G., 2013, MNRAS, 430, Sellwood J. A., Wilkinson A., 1993, Rep. Prog. Phys., 56, 173 Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021 472 Shaw R. A., Dufour R. J., 1995, PASP, 107, 896 Lopez-S´ anchez´ A.´ R., Esteban C., 2010, A&A, 517, A85 Shields G. A., Skillman E. D., Kennicutt Robert C. J., 1991, ApJ, 371, 82 Lopez-S´ anchez´ A.´ R., Westmeier T., Esteban C., Koribalski B. S., 2015, Skillman E. D., 1985, ApJ, 290, 449 MNRAS, 450, 3381 Skillman E. D., Kennicutt Robert C. J., Shields G. A., Zaritsky D., 1996, ApJ, Magrini L., Stanghellini L., Corbelli E., Galli D., Villaver E., 2010, A&A, 462, 147 512, A63 Smith H. E., 1975, ApJ, 199, 591 Makarov D., Prugniel P., Terekhova N., Courtois H., Vauglin I., 2014, A&A, Sofue Y., 2018, PASJ, 70, 31 570, 13 Spitoni E., Matteucci F., 2011, A&A, 531, A72 Mao Y.-W., Lin L., Kong X., 2018, ApJ, 853, 151 Stanghellini L., Magrini L., Casasola V., 2015, ApJ, 812, 39 Marino R. A. et al., 2013, A&A, 559, A114 Stanghellini L., Magrini L., Villaver E., Galli D., 2010, A&A, 521, A3 Marinova I., Jogee S., 2007, ApJ, 659, 1176 Stasinska´ G., 1982, A&AS, 48, 299 Martinez-Valpuesta I., Gerhard O., 2011, ApJ, 734, L20 Stauffer J. R., Bothun G. D., 1984, AJ, 89, 1702 Martin P., 1995, AJ, 109, 2428 Storchi-Bergmann T., Wilson A. S., Baldwin J. A., 1996, ApJ, 460, 252 Martin P., Roy J.-R., 1994, ApJ, 424, 599 Sanchez-Menguiano´ L. et al., 2016, A&A, 587, A70 Massey P., Olsen K. A. G., Hodge P. W., Strong S. B., Jacoby G. H., Sanchez-Menguiano´ L. et al., 2018, A&A, 609, A119 Schlingman W., Smith R. C., 2006, AJ, 131, 2478 Sanchez´ S. F. et al., 2012, A&A, 546, A2 McCall M. L., Rybski P. M., Shields G. A., 1985, ApJS, 57, 1 Sanchez´ S. F. et al., 2014, A&A, 563, A49 McGaugh S. S., 1991, ApJ, 380, 140 Toribio San Cipriano L., Garc´ıa-Rojas J., Esteban C., Bresolin F., Peimbert Meurer G. R. et al., 2006, ApJS, 165, 307 M., 2016, MNRAS, 458, 1866 Moustakas J., Kennicutt R. C. J., Tremonti C. A., Dale D. A., Smith J.-D. T., Torres-Peimbert S., Peimbert M., Fierro J., 1989, ApJ, 345, 186 Calzetti D., 2010, ApJS, 190, 233 Vale Asari N., Couto G. S., Cid Fernandes R., Stasinska´ G., de Amorim A. Munoz-Mateos˜ J. C. et al., 2013, ApJ, 771, 59 L., Ruschel-Dutra D., Werle A., Florido T. Z., 2019, MNRAS, 489, 4721 Mendez-Abreu´ J., Costantin L., Aguerri J. A. L., de Lorenzo-Caceres´ A., van der Walt S., Colbert S. C., Varoquaux G., 2011, Computing in Science & Corsini E. M., 2018, MNRAS, 479, 4172 Engineering, 13, 22 Nair P. B., Abraham R. G., 2010, ApJS, 186, 427 van Zee L., Salzer J. J., Haynes M. P. J. J., Haynes M. P., O’Donoghue A. A., Oey M. S., Kennicutt R. C. J., 1993, ApJ, 411, 137 Balonek T. J., 1998, AJ, 116, 2805 Oey M. S., Parker J. S., Mikles V. J., Zhang X., 2003, AJ, 126, 2317 Vila-Costas M. B., Edmunds M. G., 1992, MNRAS, 259, 121 Oey M. S. et al., 2007, ApJ, 661, 801 Vilchez J. M., Edmunds M. G., Pagel B. E. J., 1988a, PASP, 100, 1428 Osterbrock D. E., Ferland G. J., 2006, Astrophysics of Gaseous Nebulae and Vilchez J. M., Pagel B. E. J., Diaz A. I., Terlevich E., Edmunds M. G., 1988b, Active Galactic Nuclei. University Science Books, Sausalito, California MNRAS, 235, 633 Pagel B. E. J., Edmunds M. G., Blackwell D. E., Chun M. S., Smith G., 1979, Veron-Cetty´ M.-P., Veron´ P., 2006, A&A, 455, 773 MNRAS, 189, 95 Walcher C. J. et al., 2014, A&A, 569, A1 Pagel B. E. J., Edmunds M. G., Smith G., 1980, MNRAS, 193, 219 Walsh J. R., Roy J. R., 1997, MNRAS, 288, 726 Patterson M. T., Walterbos R. A. M., Kennicutt R. C., Chiappini C., Thilker Webster B. L., Smith M. G., 1983, MNRAS, 204, 743 D. A., 2012, MNRAS, 422, 401 Wegg C., Gerhard O., Portail M., 2015, MNRAS, 450, 4050 Peimbert M., Costero R., 1969, Bol. Obs. Tonantzintla Tacubaya, 5, 3 Werk J. K., Putman M. E., Meurer G. R., Santiago-Figueroa N., 2011, ApJ, Pettini M., Pagel B. E. J., 2004, MNRAS, 348, L59 735, 71 Pilyugin L. S., Grebel E. K., 2016, MNRAS, 457, 3678 Wozniak H., Friedli D., Martinet L., Martin P., Bratschi P., 1995, A&AS, 111, Pilyugin L. S., Grebel E. K., Kniazev A. Y., 2014, AJ, 147, 131 115 Pilyugin L. S., Grebel E. K., Zinchenko I. A., Nefedyev Y. A., V´ılchez J. M., YinS.Y.,LiangY.C.,HammerF.,BrinchmannJ.,ZhangB.,DengL.C., 2019, A&A, 623, A122 Flores H., 2007, A&A, 462, 535 Perez-Montero´ E., 2014, MNRAS, 441, 2663 York D. G. et al., 2000, AJ, 120, 1579 Perez-Montero´ E., 2017, PASP, 129, 043001 Zahid H. J., Bresolin F., 2011, AJ, 141, 192 Perez-Montero´ E., Contini T., 2009, MNRAS, 398, 949 Zanmar´ Sanchez´ R., Sellwood J. A., Weiner B. J., Williams T. B., 2008, ApJ, Perez-Montero´ E., D´ıaz A. I., 2005, MNRAS, 361, 1063 674, 797 Perez-Montero´ E. et al., 2016, A&A, 595, A62 Zaritsky D., Kennicutt Robert C. J., Huchra J. P., 1994, ApJ, 420, 87 Relano˜ M., Monreal-Ibero A., V´ılchez J. M., Kennicutt R. C., 2010, MNRAS, Zaritsky D., Rix H.-W., Rieke M., 1993, Nature, 364, 313 402, 1635 Zhang K. et al., 2017, MNRAS, 466, 3217 Robertson P., Shields G. A., Blanc G. A., 2012, ApJ, 748, 48 Zinchenko I. A., Just A., Pilyugin L. S., Lara-Lopez M. A., 2019, A&A, 623, Robitaille T., Bressert E., 2012, Astrophysics Source Code Library, A7 ascl:1208.017 Zou Y., Shen J., Li Z.-Y., 2014, ApJ, 791, 11 Rosales-Ortega F. F., D´ıaz A. I., Kennicutt R. C., Sanchez´ S. F., 2011, Zurita A., Bresolin F., 2012, MNRAS, 427, 1463 MNRAS, 415, 2439 Zurita A., Florido E., Bresolin F., Perez´ I., Perez-Montero´ E., 2020, MNRAS Rosolowsky E., Simon J. D., 2008, ApJ, 675, 1213 in press (Paper II) Roy J. R., Walsh J. R., 1997, MNRAS, 288, 715 Zurita A., Rozas M., Beckman J. E., 2000, A&A, 363, 9

MNRAS 500, 2359–2379 (2021) Chemical abundances and gradients in spirals 2379

SUPPORTING INFORMATION Any queries (other than missing material) should be directed to the corresponding author for the article. Supplementary data are available at MNRAS online. bars I accepted APPENDICES.pdf.

Please note: Oxford University Press is not responsible for the content This paper has been typeset from a TEX/LATEX file prepared by the author. or functionality of any supporting materials supplied by the authors. Downloaded from https://academic.oup.com/mnras/article/500/2/2359/5893801 by Inst. Astrofisica Andalucia CSIC user on 26 April 2021

MNRAS 500, 2359–2379 (2021)