The Frequency of Dust Lanes in Edge-On Spiral Galaxies Identified by Galaxy Zoo in Kids Imaging of GAMA Targets

Home , Dust lane

University of Louisville ThinkIR: The University of Louisville's Institutional Repository

Faculty Scholarship

1-1-2019

The frequency of dust lanes in edge-on spiral galaxies identified by galaxy Zoo in KiDS Imaging of GAMA Targets

Benne W. Holwerda University of Louisville, [email protected]

Lee Kelvin Liverpool John Moores University

Ivan Baldry Liverpool John Moores University

Chris Lintott

Mehmet Alpaslan New York University

See next page for additional authors

Follow this and additional works at: https://ir.library.louisville.edu/faculty

Part of the Astrophysics and Astronomy Commons

ThinkIR Citation Holwerda, Benne W.; Kelvin, Lee; Baldry, Ivan; Lintott, Chris; Alpaslan, Mehmet; Pimbblet, Kevin A.; Liske, Jochen; Kitching, Thomas; Bamford, Steven; De Jong, Jelte; Bilicki, Maciej; Hopkins, Andrew; Bridge, Joanna; Steele, R.; Jacques, A.; Goswami, S.; Kusmic, S.; Roemer, W.; Kruk, S.; Popescu, C. C.; Kuijken, K.; Wang, L.; and Wright, A., "The frequency of dust lanes in edge-on spiral galaxies identified yb galaxy Zoo in KiDS Imaging of GAMA Targets" (2019). Faculty Scholarship. 490. https://ir.library.louisville.edu/faculty/490

This Article is brought to you for free and open access by ThinkIR: The University of Louisville's Institutional Repository. It has been accepted for inclusion in Faculty Scholarship by an authorized administrator of ThinkIR: The University of Louisville's Institutional Repository. For more information, please contact [email protected]. Authors Benne W. Holwerda, Lee Kelvin, Ivan Baldry, Chris Lintott, Mehmet Alpaslan, Kevin A. Pimbblet, Jochen Liske, Thomas Kitching, Steven Bamford, Jelte De Jong, Maciej Bilicki, Andrew Hopkins, Joanna Bridge, R. Steele, A. Jacques, S. Goswami, S. Kusmic, W. Roemer, S. Kruk, C. C. Popescu, K. Kuijken, L. Wang, and A. Wright

This article is available at ThinkIR: The University of Louisville's Institutional Repository: https://ir.library.louisville.edu/ faculty/490 Draft version September 18, 2019 Typeset using LATEX twocolumn style in AASTeX62

The frequency of dust lanes in edge-on spiral galaxies identiﬁed by Galaxy Zoo in KiDS imaging of GAMA targets

Benne W. Holwerda,1 Lee Kelvin,2 Ivan Baldry,2 Chris Lintott,3 Mehmet Alpaslan,4 Kevin A Pimbblet,5 Jochen Liske,6 Thomas Kitching,7 Steven Bamford,8 Jelte de Jong,9 Maciej Bilicki,9 Andrew Hopkins,10 Joanna Bridge,1 R. Steele,1 A. Jacques,1 S. Goswami,1 S. Kusmic,1 W. Roemer,1 S. Kruk,11 C.C. Popescu,12, 13 K. Kuijken,9 L. Wang,14, 15 and A. Wright16

1Department of Physics and Astronomy, 102 Natural Science Building, University of Louisville, Louisville KY 40292, USA 2Astrophysics Research Institute, Liverpool John Moores University, IC2, Liverpool Science Park, 146 Brownlow Hill, Liverpool, L3 5RF, United Kingdom 3Denys Wilkinson Building Keble Road Oxford OX1 3RH, United Kingdom 4Center for Cosmology and Particle Physics Department of Physics, New York University 726 Broadway, Office 913 New York, NY 10003, USA 5E.A.Milne Centre for Astrophysics, University of Hull, Cottingham Road, Kingston-upon-Hull, HU6 7RX, UK 6UniversitätHamburg, Hamburger Sternwarte, Gojenbergsweg 112, 21029 Hamburg, Germany 7Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, Surrey RH5 6NT, UK 8University of Nottingham, University Park, Nottingham, NG7 2RD, United Kingdom 9Leiden Observatory, Universiteit Leiden, Niels Bohrweg 2,NL-2333 CA Leiden, The Netherlands 10Australian Astronomical Observatory 105 Delhi Rd, North Ryde, NSW 2113, Australia 11University of Oxford, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, United Kingdom 12Jeremiah Horrocks Institute, University of Central Lancashire, Preston, United Kingdom 13The Astronomical Institute of the Romanian Academy, Str. Cutitul de Argint 5, Bucharest, Romania 14SRON Netherlands Institute for Space Research, Landleven 12, 9747 AD, Groningen, The Netherlands 15Kapteyn Astronomical Institute, University of Groningen, Postbus 800, 9700 AV, Groningen, The Netherlands 16Argelander-Institut fürAstronomie (AIfA) UniversitätBonn Auf dem Hügel 71 D-53121 Bonn, Germany

(Received January 1, 2018; Revised January 7, 2018; Accepted September 18, 2019) Submitted to ApJ

ABSTRACT Dust lanes bisect the plane of a typical edge-on spiral galaxy as a dark optical absorption feature. Their appearance is linked to the gravitational stability of spiral disks; the fraction of edge-on galaxies that displays a dust lane is a direct indicator of the typical vertical balance between gravity and turbulence; a balance struck between the energy input from star-formation and the gravitational pull into the plane of the disk. Based on morphological classifications by the Galaxy Zoo project on the Kilo-Degree Survey (KiDS) imaging data in the Galaxy and Mass Assembly (GAMA) fields, we explore the relation of dust lanes to the galaxy characteristics, most of which were determined using the magphys spectral energy distribution fitting tool: stellar mass, total and specific star-formation rates, and several parameters describing the cold dust component. arXiv:1909.07461v1 [astro-ph.GA] 16 Sep 2019 We find that the fraction of dust lanes does depend on the stellar mass of the galaxy; they start to ∗ 9 5 appear at M ∼ 10 M . A dust lane also implies strongly a dust mass of at least 10 M , but otherwise does not correlate with cold dust mass parameters of the magphys spectral energy distribution analysis, nor is there a link with star-formation rate, specific or total. Dust lane identification does not depend on disk ellipticity (disk thickness) or Sersic profile but correlates with bulge morphology; a round bulge favors dust lane votes.

Corresponding author: Benne W. Holwerda [email protected] 2 Holwerda et al.

The central component along the line of sight that produces the dust lane is not associated with either one of the components ﬁt by magphys, the cold diﬀuse component or the localized, heated component in HII regions, but a mix of these two.

Keywords: editorials, notices — miscellaneous — catalogs — surveys

1. INTRODUCTION the dust morphology and any transition in structure. A dark stripe in the mid-plane of the spiral disk is However, the height of the disk is only just resolved at part of the canonical edge-on view of late-type galax- the longer wavelengths, and the emission depends on ies. These dust lanes are so common that their pres- the temperature of the cold dust grains. It has proven ence is often taken as a signature of a perfectly edge-on difficult to disentangle the vertical ISM density profile disk (inclination i > 85◦). Dalcanton et al.(2004) show from the vertical temperature gradient. that dust lanes appear predominantly in massive galax- The thickness of the dusty ISM disks was shown first 9.8 by the Radiative Transfer (RT) models of several disks ies (vrot > 120 km/s or a stellar mass of ∼ 10 M ). They link the phenomenon to the vertical stability – the by Xilouris et al.(1999). Alton et al.(1998, 2000) fol- Toomre Q criterion (Toomre 1964) – of the gas and stel- low the initial results with more NGC 891 observations lar spiral disk that hosts the dust lanes: if the surface that show a large fraction of the disk has dust emission density is sufficiently high, the disk vertically collapses associated with it. into a thin disk. In smaller galaxies, the interstellar Building on these initial result, several groups have matter (ISM) is relatively more distributed throughout constructed RT models to explain multi-wavelength the height of the stellar disk, i.e., the amount of dusty data of edge-on galaxies. There are several groups us- ISM is the same relative to the stellar mass but is not ing RT models (mostly of NGC 891) to map typical concentrated in the plane to form the line-of-sight dust dust in spiral galaxies. Popescu et al.(2000, 2011) lane seen in the edge-on disk. However, the Dalcan- model NGC 891 in detail with a diffuse disk and stel- ton et al.(2004) sample is small (49 galaxies) and is lar nursery components to explain its multi-wavelength made up of predominantly bulge-less galaxies. Obric behavior. Misiriotis et al.(2001) follows this up with 4 et al.(2006) did an initial pass on the SDSS galaxies more galaxies. This NGC 891 model is the basis for a and found that the fraction of dust lanes dramatically correction of disk galaxy photometry etc (Pastrav et al. 2012; Wijesinghe et al. 2011a,b; Grootes et al. 2013). increased at vrot = 150 km/s for all late-types. Both these studies point to a fundamental change in spiral Bianchi(2008) present the TRADING RT code and disks with halo or stellar mass. At a critical halo size, model NGC 891 with it (Bianchi & Xilouris 2011) as the disk flattens conspicuously with respect to it’s size well as follow-up on the initial Xilouris work (Bianchi – the ISM more so than the stellar disk. This has im- 2007). Schechtman-Rook et al.(2012); Schechtman- plications for the observed global galaxy characteristics: Rook & Bershady(2013) analyze NGC 891 in a com- a condensed ISM disk may form stars more efficiently, pletely new way, using large-scale SED models together the vertical instability affects the spiral density wave, with small-scale dust structures observed in HST fields the formation efficiency of bars may change, a compact to model the dust disk of this canonical edge-on galaxy. dusty ISM lowers the UV photon escape fraction (e.g. The current state-of-the-art in edge-on galaxy radia- Dijkstra & Wyithe 2012; Stark et al. 2015; Dijkstra et al. tive transfer is the SKIRT fitting code Baes & Dejonghe 2014, 2016; Bridge et al. 2018). If the transition is in- (2001a,b); Baes et al.(2011). The power of this model- deed sudden, it constitutes a fundamental phase change ing code is that is has a fitting component (De Geyter in the ISM of spirals. et al. 2013), making it by far the most suitable for fu- In emission, the picture should be clearer: there is no ture work and flexible enough to quickly incorporate new need for a stellar disk to backlight the dust structures. data. SKIRT has resulted in models that have come Thus, sub-mm observations with Herschel and Spitzer of closest to explaining the full multi-wavelength data on vertically resolved, edge-on disks should reveal if there Milky-Way type edge-on galaxies (HEROES Herschel does exist a sharp transition in the dusty ISM structure. project de Looze et al. 2012; Verstappen et al. 2013; Al- Several programs with Herschel target massive edge-on laert et al. 2015; De Geyter et al. 2014, 2015; Mosenkov spirals, notably the HEROES project (Verstappen et al. et al. 2016). Once again NGC 891 makes an appearance 2013). The NHEMESES program (Holwerda et al. 2011; in the first SKIRT efforts (Hughes et al. 2014, 2015). Holwerda et al. 2012b, Holwerda et al. in preparation) The main issue for SKIRT remains an under-prediction is designed specifically to target smaller disks to explore of the face-on optical depth (e.g. Holwerda 2005) or an Dust Lanes In Edge-On Galaxies in GAMA/KiDS 3 under-prediction of the sub-mm emission (Saftly et al. 8 0.14 2015) ) The main –generalized– results coming out of these r 9 0.12 y SKIRT efforts for massive spirals is that the dust is in a / 1 0.10

( 10

disk with a scale-height 50% of the stellar disk’s scale- 0 height and the dust disk’s scale-length is 150% the scale- 1 0.08 g 11 o length of the stellar disk respectively, initially already l Redshift (z) 12 0.06 reported in Xilouris et al.(1999). In addition to this dif- R fuse disk, clumpy structures are around star-formation F 0.04 S 13 regions. This model of diffuse+clumps is used in the S 0.02 magphys SED code (da Cunha et al. 2012) as well. The 14 balance between the small and large dust structure mod- 6 8 10 els is key to furthering our understanding of their stellar Stellar Mass log10(M * /M ) light from this point (Saftly et al. 2015). A big strength is that optical measurements can be compared directly Figure 1. The stellar mass and specific star-formation to dust emission (e.g., Hughes et al. 2015). plot of the GAMA galaxies classified by GalaxyZoo with the A smaller effort is under way to characterize the GAMA spectroscopic redshift for the colorbar. The detection disk galaxies much less massive than the Milky Way and inclusion of low mass galaxies are biased towards high (NHEMESES Holwerda et al. 2011,& in prep). A re- specific star-formation and low redshift. maining issue with SKIRT and all the other RT models is that the face-on central optical depth remains low in necessarily a non-transient one – i.e., dust lanes may still comparison with transmission measurements by a factor be rapidly both destroyed and re-formed. However, the ∼ 2 (Holwerda 2005) and (Keel et al. 2013). edge-on view, which is often optically thick, is the most A complementary effort is therefore to leverage the robust to such changes. In this paper, we explore the statistics of optical imaging surveys on dust lane fre- links in the local Universe between dust lane occurrence quency. We use data from the citizen science project and disk properties. Galaxy Zoo1 (Lintott et al. 2008a) for better statistics Our goals for this paper are to explore (a) if the sharp of absorption features in late-type edge-on galaxies. transition in dust lane frequency is still seen at the The Galaxy Zoo project has already proven itself un- stellar mass that Dalcanton et al.(2004) observed in paralleled in the large-scale analysis of morphological bulgeless galaxies, (b) what the effect of a bulge is on phenomena, until recently the purview of specialist clas- dust lane frequency, and (c) the relation between galaxy sifiers. For example, Galaxy Zoo classifications (Lintott properties and dust lane frequency. This paper is orga- et al. 2008b, 2011; Fortson et al. 2012) have been used nized as follows: §2 describes the sample selection from to identify mergers (Darg et al. 2010a,b, 2011; Casteels the Galaxy Zoo database, §3 describes the part of the et al. 2013), the prevalence of bars (Hoyle et al. 2011; GAMA-KiDS Galaxy Zoo decision tree relevant to this Masters et al. 2011, 2012; Kruk et al. 2018), and occult- project, §4 presents the results for the numbers of galax- ing galaxy pairs (Keel et al. 2013). The identification ies identified as disk, edge-on and with a dust lane as a of dusty structures in SDSS images has already proven function of various galaxy properties, §5 briefly discusses very scientifically worthwhile: Kaviraj et al.(2012a) and these results and §6 lists our conclusions. Shabala et al.(2012) show how dust structures prevail in massive elliptical galaxies. Here we focus on those 2. SAMPLE SELECTION AND DATA galaxies identified by the Galaxy Zoo as disk-dominated, The Galaxy Zoo classifications are based on the spiral galaxies, seen edge-on. Galaxy and Mass Assembly survey DR2 and the Kilo Holwerda et al.(2012a) show that the dust lane frac- Degree Survey (KiDS) imaging. For the Galaxy Zoo tion in massive L? galaxies barely changes with redshift V classification, 49851 galaxies were selected from the out to z ∼ 0.8. This was calibrated with a select sample equatorial fields with redshifts z < 0.15. The Galaxy of L? SDSS galaxies, for which they also found a dust V Zoo provided a monumental effort with almost 2 million lane fraction of ∼ 80%. This indicates that the dust lane classifications received from over 20,000 unique users is a very constant phenomenon in massive disks, if not over the course of 12 months. The Kilo Degree Survey (KiDS de Jong et al. 2013, 1 http://www.GalaxyZoo.org 2015, 2017) is an ongoing optical wide-field imaging survey with the OmegaCAM camera at the VLT Survey 4 Holwerda et al.

2 Telescope. It aims to image 1350 deg in four ﬁlters (u 104 g r i). The core science driver is mapping the large-scale matter distribution in the Universe, using weak lensing 103 shear and photometric redshift measurements. Further science cases include galaxy evolution, Milky Way struc- 102 ture, detection of high-redshift clusters, and ﬁnding rare sources such as strong lenses and quasars. KiDS image 101 quality is typically 000. 6 resolution (for sdss-r) and depths of 23.5, 25, 25.2, 24.2 magnitude for i, r, g and u respec- 100 All tively. Edge-on

GAMA is a combined spectroscopic and multi- Number of Galaxies Dust Lane 10 1 wavelength imaging survey designed to study spatial 0.0 0.2 0.4 0.6 0.8 1.0 structure in the nearby (z < 0.25) Universe on kpc Number of votes for Spiral to Mpc scales (see Driver et al. 2009, 2011, for an overview). The survey, after completion of phase 2 Figure 2. The number GAMA/KiDS galaxies as a function (Liske et al. 2015), consists of three equatorial regions of the fraction of votes in favor of galaxy with features (A01 each spanning 5 deg in Dec and 12 deg in RA, centered in Figure3), edge-on (T01 in Figure3), and a dustlane (T06 in RA at approximately 9h (G09), 12h (G12) and 14.5h in Figure3). The solid line are all the objects and the frac- (G15) and two Southern fields, at 05h (G05) and 23h tion of votes in favor of a galaxy with features, the dashed line is a subset of these (f > 0.5) and the fraction of (G23). The three equatorial regions, amounting to a features votes in favor of the galaxy being edge-on. The dotted line total sky area of 180 deg2, were selected for this study. is a subset of these (ffeatures > 0.5 and fedge−on > 0.5) with For the purpose of visual classification, 49851 galaxies the fraction of votes in favor of a dust lane. were selected from the equatorial fields with redshifts z < 0.15. Figure2 shows the distribution of votes for data. In addition to the magphys data, use the Sèrsic galaxies in our subsample of disk galaxies (T00 in Fig- fits to the UKIDSS (Kelvin et al. 2014). ure3 question has been answered by more than 50% Figure1 shows the magphys stellar mass and specific of the volunteers as a disk galaxy). The GAMA survey star-formation rate plot with the redshift indicated as is >98% redshift complete to r < 19.8 mag in all three well. One can discern selection effects e.g. how lower- equatorial regions. We use two data-products described mass objects are only found at lower redshifts and more in the third GAMA data-release (DR3, Baldry et al. massive objects can be found out to the highest red- 2018): the magphys SED fits (Driver et al. 2018) and shifts. the Sèrsicfit catalogs (Kelvin et al. 2014). The GAMA-KiDS Galaxy Zoo project uses the de- 3. GALAXY ZOO DECISION TREE cision tree in use for the latest (4th) iterations of the Figure3 shows the decision tree for the Galaxy Zoo Zoo. KiDS cutouts were introduced to the classifica- fourth iteration, of which the GAMA/KiDS classification pool and mixed in with the ongoing classification tions are part. After the initial decision if a galaxy is efforts. Scientific aims include correlating general mor- smooth, disk, or star/artifact (T00), second tier ques- phology to the GAMA results using the full suite of tions are asked of the volunteers (T07 or T01). In the multi-wavelength and spectral information and the iden- case of disk galaxies, the follow up question determines tification of rare features (e.g. strong lensing arcs of if the disk is viewed edge-on or not (T01). This is a galaxy occultation). A full description of the GAMA- critical question for this project as we are interested in KiDS Galaxy Zoo effort can be found in Kelvin et al. in the prevalence of dust lanes in edge-on disks. The next preparation. follow up is to determine what kind of bulge is visi- In addition to the GAMA-KiDS Galaxy Zoo classifi- ble: none, a box-shaped or round one (T08). The next cations, we use the magphys (da Cunha et al. 2012), tier question is whether any signs of interaction with a spectral energy distribution fits to the GAMA multi- nearby galaxy is evident (T05). Finally, the last ques- wavelength photometry (Wright et al. 2017), presented tion is a series of morphological features (ring, lens arc, in Driver et al.(2018). magphys computes stellar dust lane, irregular, other or overlap). More than one mass, specific star-formation rate, dust mass, cold dust choice can be marked. Figure2 shows the distribution fraction, cold dust temperature, average face-on optical of votes in favor of disk galaxies (T00), edge-on (T01), depth of each galaxy, which we can compare against the and displaying a dust lanes (T06). To select a galaxy as dust lane identifications in the GAMA-KiDS Galaxy Zoo having a feature, we require 50% of the votes in favor Dust Lanes In Edge-On Galaxies in GAMA/KiDS 5

T00: Is the galaxy simply smooth and rounded, with no sign of a disk? A0: Smooth A1: Features A2: Star or or disk artifact

T07: How rounded is it? T01: Could this be a disk viewed edge-on? A0: A1: In A2: Cigar A0: Yes A1: No Completely between shaped round

T08: Does the galaxy have a bulge T02: Is there a sign of a bar feature through the at its centre? If so, what shape? centre of the galaxy? A0: A1: Boxy A2: No A0: Bar A1: No bar Rounded bulge

T03: Is there any sign of a spiral arm pattern? A0: Spiral A1: No spiral

T09: How tightly wound do the spiral arms appear? A0: Tight A1: Medium A2: Loose

T10: How many spiral arms are there? A0: 1 A1: 2 A2: 3 A3: 4 A4: More than 4

T04: How prominent is the central bulge, compared with the rest of the galaxy? A0: No A2: Obvious A3: bulge Dominant

T05: Is the galaxy currently merging or is there any sign of tidal debris? A0: Merging A1: Tidal A2: Both A3: Neither debris

1st Tier Question 2nd Tier Question 3rd Tier Question T06: Do you see any of these odd features in the image? 4th Tier Question A0: None A1: Ring A2: Lens or A3: Dust A4: Irregular A5: Other A6: arc lane Overlapping

End

Figure 3. The decision tree of the Galaxy Zoo iteration 4, which was followed by the GAMA/KiDS GZ iteration. 6 Holwerda et al.

1.0 inclinations than other selection methods (e.g. near- 8 infrared ellipticity) because we wanted to compare the ) r 9 0.8 results to these galaxy properties. For example, we want y / to know the eﬀects of a substantial bulge and an ellip- 1

( 10 0.6 ticity selection, as has been typically done before, would 0 1 bias against early (Sa etc.) types.

g 11

o The term “edge-on” is somewhat subjective. In l 0.4 12 Galaxy Zoo 2, using SDSS images, about 20% of Willett R

F et al.(2013) of disk galaxies are considered “edge-on”,

S 13 0.2

S using a 70% votes in favor. Here we use a looser frac- Fraction of Edge-on Identifications 14 tion (50%) but this may well affect the final fraction of 0.0 galaxies identified with a dust lane as well. 6 8 10 Stellar Mass log10(M * /M ) 8 0.6

Figure 4. The relation between stellar mass from the ) magphys ﬁt, and star-formation with the fraction of votes r 9 y in favor of an edge-on disk. Galaxies voted in favor of edge- / 0.5 1 10 on disk are spread throughout star-formation rate and stellar ( 0 0.4 mass. 1 11 g

o 0.3 l of a disk or edge-on (fdisk >50% or fedgeon >50%) and 12 R 0.2 10% for the identiﬁcation of a dust lane (fdustlane >10%) F

S 13 because votes for them are rare (Figure2). The selec- S 0.1 tion threshold for dustlanes is set to be inclusive be- 14 Fraction of Dust Lane Identifications cause other morphological features are rarely mistaken 0.0 8 10 for dust lanes and dust lanes are so commonly thought of as an edge-on “normal” feature that they are not re- Stellar Mass log10(M * /M ) marked upon. Voting fractions have been de-biased us- Figure 5. The relation between stellar mass and star- ing the now standard Galaxy Zoo calibrations of votes formation from the magphys fit, with the fraction of votes (see Kelvin et al. in preparation Hart et al. 2016). identifying a dust lane in galaxies with > 50% of the vote in The improvement over the original Galaxy Zoo is that favor of edge-on disk. these questions are not behind a gate question of “is there anything odd?”. Many users considered dust lanes not odd and would therefore not choose the “odd” but- 3.2. Dust Lane Identification ton. The remaining issue is that votes for one morphological features may draw away votes from another. Figure5 shows the fraction of dust-lane votes in the Nevertheless, we assume all these features are relatively stellar mass and specific star-formation plot. Dust lanes rare enough for this to be not too great an issue. votes occur throughout the stellar mass and specific star- formation. Because there are many options to choose from in the final question, we only require 10% of the 3.1. Edge-on Disk Identification votes for us to consider the galaxy to have a dust lane. We select edge-ons by requiring half of the votes by Requiring a higher fraction leads to similar results but the volunteers in favor of the edge-on question. Figure with lower statistical confidence due to a smaller sample. 4 shows the number of edge-on votes in the stellar mass Our reasoning is that dust lanes are not very remarkable and specific star-formation plot. Dust lanes in thicker so a few votes in favor means it is clearly present. Figure edge-on disks (more massive galaxies) can be identified 6 shows a few randomly drawn examples of the edge-on out to greater distances (Figure1). There is therefore sample with different fractions of votes in favor of a dust an unavoidable bias in our sample against more distant, lane. We caution against using examples such as these low-mass galaxies, both in the GAMA/KiDS Galaxy to draw a criterion; Figure6 is purely for illustrative Zoo survey and the edge-on identification in the KiDS purposes. images. At the maximum distance of z=0.14, the KiDS nom- We opted for Galaxy Zoo identification of edge-on inal resolution (000. 6) corresponds to ∼ 1.6 kpc. Only disks even though it allows for many more possible disk in the most massive disk systems would a dust lane Dust Lanes In Edge-On Galaxies in GAMA/KiDS 7

Figure 6. Randomly selected examples with fedgeon > 0.5 from the GalaxyZoo GAMA sample in order of dust lane vote fraction. Starting at the top left with fdustlane = 0 and ending with fdustlane = 0.9bottom right. stand out at this resolution (e.g. NGC 891). The 4.1.1. Stellar Mass GAMA/KiDS Galaxy Zoo sample is less complete at the low-mass end with a maximum redshift for the low-mass Figure7 shows the distribution of stellar mass, as de- ∗ 9 (M < 10 M ) galaxies at z ∼ 0.04, corresponding to a termined by magphys in our sample, both as a his- linear resolution of half a kiloparsec, enough to identify togram and a fraction of all the disk-identified galax- a dust lane. ies with errors calculated using the prescription from Dust lane identification is therefore incomplete at the Cameron(2011). Edge-on disk galaxies follow the full high-mass end (part of the sample is too far away for sample of disk galaxies very well. We note a com- positive identification in all cases) and incomplete at plete absence of edge-on galaxies with a dust lane below 9 the low-mass end due to survey volume. Over the entire 10 M . Galaxies with masses below this limit are only sample of edge-on galaxies (identified as such by the included in the GAMA/Galaxy Zoo with redshifts below Galaxy Zoo) about 50 per cent of the edge-on galaxies z ∼ 0.04. On a linear scale, the KiDS resolution (000. 6) display a dust lane. Dalcanton et al.(2004); Obric et al. translates approximately 0.5 kpc at this distance. Thus (2006) and Holwerda et al.(2012b) find that the dust any clear dustlanes in the low-mass galaxies should be lane fraction lies around 80% for massive disks. identifiable. We note that there are inevitable biases introduced Several selection effects in the identification of both by the Galaxy Zoo voted selection for edge-ons: inclu- edge-ons and dust lanes may well play a role here. The sion of much more earlier type galaxies, not necessarily GAMA survey is the complete for the lower-mass edge- disk dominated ones and galaxies not perfectly edge-on. ons in the smallest volume. The small number statistics These will inevitably lower the overall number of galax- at the lower end means that if a few low-mass edge-on ies with a dust lane overall (see the discussion in Kaviraj galaxies with a dust lane have been misidentified, a sim- et al. 2012b). ilar fraction as the higher mass bins could still be true. The Galaxy Zoo identification scheme may well classify 4. RESULTS lower-mass edge-on spirals as ”smooth” as several exam- We plot the fraction of the galaxies along a property ples of this class show few distinguishing features, which (e.g. stellar mass or dust temperature) with the full includes features such as dust lanes. In this latter sce- disk galaxy sample, those considered edge-on, and fi- nario, the number of low-mass edge-on disks would go nally those considered edge-on with a dust lane identi- up but unlikely that the fraction of low-mass edge-on fied. Uncertainties are calculated from the parent sam- galaxies with a dust lane would increase. ple and the fraction identified using the prescription in The trend in Figure7 with mass is consistent with Cameron(2011). the result from Dalcanton et al.(2004), who noted a distinct changeover below and above 120 km/s rotation 4.1. magphys Stellar Mass and (Specific) speed (M ∗ ∼ 109.8M ). Dalcanton et al.(2004) and Star-formation Rate Holwerda et al.(2012b) find that 80% of the massive 8 Holwerda et al.

All 1.0 Edge-on EdgeOns Dust Lane 102 Dust Lane 0.8

0.6 101

0.4 100 0.2 Number of Galaxies 10 1 Fraction of Galaxies 0.0 6 8 10 12 4 3 2 1 0 1 Stellar Mass log (M * /M ) 10 SFR log10(M /yr)

1.0 Edge-on Figure 8. The fraction of galaxies marked as edge-on Dust Lane (dashed lines) or edge-on with a dust lane (dotted line) as a 0.8 function of the star-formation rate of all the galaxies. Gray shaded areas are the uncertainties in both fractions. 0.6

0.4 1.0 Edge-on Dust Lane 0.2 0.8

Fraction of Galaxies 0.0 0.6 7 8 9 10 * Stellar Mass log10(M /M ) 0.4

Figure 7. The histogram (top) of the stellar mass of all 0.2 the galaxies voted as disks (solid line), edge-on (dashed line) and with a dust lane (dotted line). The fraction of galaxies Fraction of Galaxies 0.0 classified as edge-on (dashed line, fedgeon >50% of the votes for selection) and the fraction that is voted both edge-on and 14 13 12 11 10 displaying a dustlane (dotted line, fedgeon > 50%, fdustlane > SSFR log10(1/yr) 10%). Gray shaded areas are the standard deviation in the numbers in each bin, both edge-on votes and edge-on and Figure 9. The fraction of galaxies as a function of the with a dust lane. Dust lanes are identified in galaxies more specific star-formation rate voted as edge-on (dashed line) 8.5 massive than 10 M , consistent with previous dust lane and with a dust lane (dotted line). Gray shaded areas are searches in the local and distant Universe (Dalcanton et al. the uncertainties in both fractions. 2004; Holwerda et al. 2012b). We will use fractions of the galaxy populations for further comparisons. at the same fraction in edge-on galaxies. Naively, one could expect the star-formation rate influencing or being disks have a dust lane. In this Galaxy Zoo sample, the influenced by the presence of a dust lane. For example, fraction lies lower however. a dark dust lane, associated with a compact molecu- The fraction of galaxies identified with a dust lane lar component of the ISM that fuels star-formation or is approximately half of those identified as edge-on but the additional turbulence thanks to newly formed stars consistent with the fraction of 80% found by previous could dissipate the compact dust lane. But no depen- authors. dency on the total star-formation rate for the prevalence of dust lanes is evident. Models for massive edge-on 4.1.2. Total Star-Formation Rate disks (Popescu et al. 2000, 2011; Popesso et al. 2012) Figure8 shows the fractions of galaxies as a function of find that the edge-on dust lane is more the line-of-sight star-formation for edge-on and dust lane identification. effect of the diffuse component of the dusty ISM rather At any given level of star-formation, dust lanes occur than associated with the star-forming dust clouds. Dust Lanes In Edge-On Galaxies in GAMA/KiDS 9

4.1.3. Specific Star-Formation Rate 1.0 Edge-on Figure9 shows the histogram for specific star- Dust Lane formation for edge-ons and those with dust lanes. At 0.8 any given level of specific star-formation, dust lanes occur at the same fraction in edge-on galaxies. Specific 0.6 star-formation is the relative growth of the stellar pop- ulation and we hoped for a better indicator of what is 0.4 the dominant mechanism rearranging the dusty ISM: gravitational contraction balanced by turbulence dis- 0.2 persing the molecular clouds throughout the height of the disk. However, like star-formation, there is no clear Fraction of Galaxies 0.0 specific star-formation rate where dust lanes become 5 6 7 8 more prevalent in edge-on galaxies. * Dust Mass log10(M /M ) 4.2. magphys Dust Output Parameters Figure 10. The fraction of disk galaxies voted edge-on Magphys outputs several parameters directly related (dashed line) and those voted disk, edge-on and showing a to the dusty ISM of a galaxy as it reprocesses star- dust lane (dotted line) as a function of magphys dust mass. light into far-infrared and sub-mm emission. magphys Dust lanes are identified throughout the range of magphys 5−6 treats the dusty ISM as a diffuse disk of colder ISM dust masses MD ∼ 10 M . with clumps of heated dust close to the ongoing star- formation. Magphys output includes the dust mass, the fraction of dust mass in the cold component, the Edge-on Dust Lane temperature of the cold component, and the average 0.8 face-on optical depth in V-band (τV ). Magphys was calibrated on local galaxies and the edge-on perspective 0.6 on disk galaxies is an edge-case to test it on, with much greater fraction of the ISM along the line-of-sight ef- 0.4 fectively opaque. Therefore, the infrared emission from the dust in the dense dust lane would be wrongly asso- 0.2 ciated to the optically thin dust at larger vertical scales. Thus the magphys output may be biased due to this mismatch in emission and attenuation effects. Fraction of Galaxies 0.0 The total dust mass or the ratio of stellar to dust mass 1 2 3 4 * are prime candidates for magphys output to correlate Stellar/Dust Mass log10(M /MD) with the presence of a dust lane in edge-on galaxies. One would naively expect for example that more dust Figure 11. The fraction of those galaxies identified as mass or relatively more dust mass would increase the disks voted as edge-on (dashed line) and those voted disk, likelihood of a dust lane if dust is distributed relatively edge-on and showing a dust lane (dotted line) as a function ∗ evenly (diffuse component) throughout a disk galaxy. of the ratio of stellar to dust mass (M /MD). The fraction of dust lanes rises gently with the stellar to dust mass ratio. 4.2.1. Dust Mass magphys reports a total dust mass for each galaxy. an inherently indirect measure of dust and reliant on Figure 10 plots the number of galaxies classified as a contrast with the surrounding stars to be noticeable, to disk, edge-on and with a dust lane as a function of dust depend on the ratio of dust to stars. mass. Dust masses in these galaxies are typically in a Figure 11 shows the fraction of edge-on and edge-on 5−6 narrow range of 10 M . Lower amounts of dust are with a dust lane as a function of stellar to dust ratio. reported for some edge-on and certainly for disk galaxies Dust lanes are identified with only a slightly increased but dust lanes occur only in a relatively narrow range frequency in low (twice as much stellar mass as dust) to of dust masses. high ratio of stellar to dust mass. 4.2.2. Star/Dust Mass Ratio 4.2.3. Cold Dust Fraction A logical follow-up is to explore if the relative masses Figure 12 shows the histogram of galaxies voted to of stars and dust. One would expect that dust lanes, have a dust lane as a function of magphys cold dust 10 Holwerda et al.

Edge-on 1.0 Edge-on Dust Lane Dust Lane 0.6 0.8

0.6 0.4 0.4 0.2 0.2 Fraction of Galaxies Fraction of Galaxies 0.0 0.0 0.0 0.2 0.4 0.6 0.8 0.0 0.5 1.0 1.5 Fraction of Cold Dust (fc) Average Optical Depth( V)

Figure 12. The fraction of galaxies identified as edge-on Figure 14. The average face-on optical depth as deter- (dashed line) and with a dust lane (dotted) as a function mined by magphys. The majority of disk galaxies is opti- of cold dust as found by magphys by Driver et al.(2018). cally thin according to magphys. The fraction of dust lanes No relation is visible: dust lane votes remain constant with declines steadily with the edge-on fraction. magphys cold dust fraction. on dust temperature is evident. The magphys cold dust temperature (T ) refers to the cold component in the 0.6 Edge-on c Dust Lane diffuse ISM. One would –perhaps naively– expect there 0.5 to be a correlation as the colder ISM would sinks to the plane of the disk as dense clumps of ISM, enhancing the 0.4 dust lane effect and the warmer ISM is distributed more 0.3 vertically. Once a cold dust component is present in the diffuse 0.2 ISM, it appears decoupled from the identification of a dust lane. 0.1 4.2.5. Optical Depth Fraction of Galaxies 0.0 16 18 20 22 24 Figure 14 shows the relation between face-on optical Temperature of Cold Dust (T ) depth computed by magphys and the fraction of dust c lane votes. The majority of galaxies in our sample are Figure 13. The fraction galaxies identified as edge-on considered optically thin by magphys. Only a few dozen (dashed line) and with dust lanes as a function of tempera- edge-on galaxies are in the optically thick regime (τV > ture of cold dust component as found by magphys by Driver 1). A dust lane is by definition optically thick and hence et al.(2018). No relation is visible; dust lane votes remain this result shows the magphys result is mostly based on constant with cold dust temperature. the light from either side of the dark, optically thick lane. Yet, given that dust lanes can be the integral fraction (fc). This is the fractional contribution by cold effect of a diffuse ISM, their presence should be related dust to the dust luminosity of the ambient ISM accord- to the inferred optical depth by magphys. The edge-on ing to the magphys best fit. There is no correlation galaxies with a dust lane remain a steady fraction of the between the fraction of cold ISM identified by magphys edge-on voted galaxies as a function of optical depth. and the fraction of votes in favor of a dust lane: a cold Optically thick galaxies present with too low statistics component can be evident in the SED fit or a dust lane to say much about their dust lane fraction. is identified in the images but the two effects do not appear to correlate at all. 4.3. Morphology Dalcanton et al.(2004) speculate that the presence of 4.2.4. Cold Dust Temperature a dust lane is also linked to the oblateness of the disk Figure 13 shows the votes for dust lanes as a function with flatter disks showing more prevalent dust lanes. In of magphys cold dust temperature (Tc). No dependence addition, one can expect that bars perturb the molecular Dust Lanes In Edge-On Galaxies in GAMA/KiDS 11 cloud arrangement that is responsible for the dark dust Dalcanton et al.(2004) make but we should bear in lane. This can be explored by comparing the dust lanes mind that both ellipticity and dust lane identification votes as a function of near-infrared (least influenced by are strongly dependent on the inclination of the disk. dust) stellar disk appearance, either disk oblateness, el- Only near perfectly edge-on will both the dust lane be lipticity, or axis ratio (B/A) or the relationship between unequivocal and the ellipticity most extreme. Because dust lane votes and votes for different bulge morpholo- we use the Galaxy Zoo edge-on identification, this result gies. may have been diluted by not perfectly edge-on systems We selected edge-on disks using the Galaxy Zoo votes with more median ellipticity values and more difficult to with the purpose of comparing them to other morpho- identify dust lanes. logical features identified and this axis ratio and ellip- A second limitation is that the optical KiDS data an- ticity explicitly to test the initial results in Dalcanton alyzed by the Galaxy Zoo is both deeper and higher et al.(2004). spatial resolution than the UKIDS K-band data, which likely results in rounder (lower ellipticity) K-band mea- 4.3.1. Bulges surements for these galaxies. The combined effects have Figure 15 shows the distribution of disk galaxies, likely smoothed out any dependence. galaxies voted as edge-on, and the number of those galaxies with a vote in favor of a dust lane as a func- 4.4. Environment tion of the fraction of votes in favor of a boxy bulge, a The question whether there are signs of an ongoing round bulge or no bulge at all. Boxy bulges are seen as interaction or merger and the presence of a dust lane evidence for a bar in edge-on galaxies. are now separated (questions Figure3). This opens the Round bulges at any level of confidence always have possibility to explore whether tidal effects of a merger the same fraction of edge-on galaxies with a dust lane. influence the presence of a dust lane. Holwerda et al. The number of galaxies with dust lanes votes anti- (2013) find that in UGC 3995, a mid-stage interaction, correlates with no bulge votes and the number of galax- the diffuse component of dust has been mostly destroyed ies with dust lane votes decreases with with increasing or swept up into dense structures. This provides a hint voter confidence for a boxy bulge. Therefore no, or that the dusty ISM is radically rearranged in the early a boxy bulge decreases the chance of a dust lane be- stages of a merger or interaction. ing identified and a round bulge has no influence on a The fraction of votes in favor of a dust lane and no dust lane identification. This result is a little counter- interaction appear to be correlated (Figure 17). Either intuitive as a prominent bulge should aid in highlight- an interaction distracts from the identification of dust ing a dust lane bisecting the plane of the disk. If boxy lanes, or dust lanes are perturbed/removed by an inter- bulges are linked to bars, as they often are in the liter- action or a combination of these effects. ature, then this points to a clearing out of dust in the inner disk, resulting in fewer dust lane identifications. 5. DISCUSSION The correlation between a lack of dust lanes with a lack of a bulge identification could be a visual selection ef- Dust lanes are common in edge-on galaxies, so much fect as dust lanes are a little less backlighted in bulgeless so that votes for their presence need not be numerous, galaxies. they are often considered unremarkable by classifiers. The presence of dust lanes depends most strongly on 4.3.2. Disk Oblateness or Ellipticity stellar mass. Dust lanes are increasingly identified where Dalcanton et al.(2004) noted how the edge-on disks the relative dust mass is smaller compared to the stellar display not only dust lanes but a flattened stellar disk disk. as well. We use the UKIDSS near-infrared ellipticity Dust lane presence depends on the type of bulge vis- to explore this observation using the Galaxy Zoo votes ible in the edge-on disk, with votes for boxy bulges for both edge-on disk and dust lanes. Figure 16 shows –thought connected to the presence of a bar– anti- the UKIDSS K-band based ellipticity and the fraction of correlated with the presence of dust lanes. Similarly, galaxies classified as edge-ons, and the fraction identified the presence of dust lanes anti-correlates with signs of with a dust lane. The numbers are remarkably steady interaction, recent or ongoing. with ellipticity, showing no preference. Similarly, the Our results point towards a scenario where dust lanes Sersic index has little influence on the prevalence of dust need unperturbed disks of a certain mass to show clearly. lanes. This is in line with the discussion by Dalcanton et al. The lack of dependence on dust lane votes on axis ra- (2004), where the vertical stability of the disk is such tio (ellipticity) is surprising given the strong rationale that dusty ISM clouds sink to the central plane as well 12 Holwerda et al.

1.0 Edge-on Edge-on 1.0 Edge-on Dust Lane 0.8 Dust Lane Dust Lane 0.8 0.8 0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

Fraction of Galaxies 0.0 Fraction of Galaxies 0.0 Fraction of Galaxies 0.0 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 Fraction of Voter for boxy bulge Fraction of Voter for round bulge Fraction of Voter for NO bulge

Figure 15. The fraction of galaxies voted on as edge-on (dashed), and displaying a dust lane (dotted line) as a function of the fraction of votes in favor of a boxy bulge (left), a round bulge (middle) and no bulge (right). The lack of a correlation with the magphys param- 0.6 Edge-on Dust Lane eters and the occurrence of dust lanes in the Galaxy 0.5 Zoo classiﬁcations is puzzling. We note that the magphys values for individual galaxies still hold large uncer- 0.4 tainties in the derived parameters (Wright et al. 2018). The thin, cold, dusty ISM responsible for the dust lane 0.3 should logically be associated with one of the two com- 0.2 ponents used in magphys; the cold, compact one, not the warm component heated by star-formation. Most 0.1 vexingly, it correlates with neither clearly.

Fraction of Galaxies 0.0 One can look to the radiative transfer results to inter- pret these result: the optically thick component in the 0.0 0.2 0.4 0.6 0.8 plane of the disk is neither warm nor cold, nor is it ex- Ellipticity (eK) clusively in dense clouds or the diﬀuse component best probed with magphys. Figure 16. The fraction of galaxies votes as edge-on The cold dusty clumps may occur in the plane of (dashed), and with a dust lane (dotted) as a function of UKIDDS ellipticity. the stellar disk, they are deeply embedded in this disk: their contribution to the line-of-sight optical depth in the edge-on perspective is smoothed out by the nearby

1.0 Edge-on parts of the stellar disk. Equally, the warmer diffuse Dust Lane component is not optically thick but it’s effect is felt 0.8 over a larger path along the line of sight, resulting in an equal contribution to the dust lane. 0.6 The stellar-to-dust mass ratio effect is similarly counter-intuitive but dust lanes need a stellar disk to 0.4 contrast against. Following this, we note that a dust lane requires a 5 0.2 minimum mass of dust MD ∼ 10 M but it can occur in 9 any stellar mass disk (above 10 M ) and any oblateness. Fraction of Galaxies 0.0 0.0 0.2 0.4 0.6 0.8 6. CONCLUSIONS Fraction of Voter for NO merger signs Using the Galaxy Zoo classifications of the KiDS data overlapping with the GAMA equatorial fields, we exam- Figure 17. The fraction of galaxies voted edge on (dashed ine the frequency of dust lanes in edge-on galaxies and line) and with a dust lane (dotted line) as a function of the relate them to other observables of the galaxies. We find fraction of votes for showing no merger signs. The fractions the following: strongly suggests that dust lanes are preferentially identified in galaxies where there no signs of a merger are found. • Dust lanes are seen to occur most frequently in 8.5 above a stellar mass of 10 M . This corresponds as the result by Holwerda et al.(2013) that shows the reasonably to the one found by Dalcanton et al. 9.8 diffuse dust is removed/swept up by an interaction in (2004) for stellar mass: 10 M corresponding to the early stages. vrot ∼ 120 km/s), (Figure7) Dust Lanes In Edge-On Galaxies in GAMA/KiDS 13

• The occurrence of a dust lane appears poorly or to discriminating whether this is a sharp transition at 9 not at all correlated with any magphys dust pa- 10 M stellar mass or a smooth one will be much im- rameters, (Figures 12, 13, 14), indicating the dust proved statistics on dust lane frequency in lower mass lane is not associated with either dust component disk galaxies. The combination of the higher resolution alone but a cumulative eﬀect of all the dust in the and statistics make that practical with WFIRST or per- disk. haps LSST.

• The dust lanes occur in galaxies with a minimum 5 dust mass of MD ∼ 10 M (Figure 10) but show a wide range of stellar to dust mass ratios (Figure ACKNOWLEDGEMENTS 11). Dust lanes may be identified more prevalently This publication has been made possible by the par- in relatively more massive stellar disks. ticipation of more than 20000 volunteers in the Galaxy Zoo project. Their contributions are individually ac- • The identification of a boxy bulge and the pres- knowledged at http://authors.GalaxyZoo.org/. ence of a dust lane appears anti-correlated (Figure This research has made use of the NASA/IPAC Extra- 15), suggesting boxy bulges (bars) are involved in galactic Database (NED) which is operated by the Jet sweeping clear their inner disk of dust. Propulsion Laboratory, California Institute of Technol- • Dust lanes and signs of interaction anti-correlate ogy, under contract with the National Aeronautics and (Figure 17), confirming a scenario where the dust Space Administration. This research has made use of ISM is rearranged early in a galaxy-galaxy inter- NASA’s Astrophysics Data System. This research made action. use of Astropy, a community-developed core Python package for Astronomy (Astropy Collaboration et al. Future work on the frequency of dust lanes in edge- 2013). This research made use of matplotlib, a Python on galaxies can employ the full analysis of the Galaxy library for publication quality graphics (Hunter 2007). Zoo classifications of the Dark Energy Sky Survey im- PyRAF is a product of the Space Telescope Science In- ages or follow-up Galaxy Zoo projects to answer ques- stitute, which is operated by AURA for NASA. This tions on the size and morphology of the dust lanes. Key research made use of SciPy (Jones et al. 2001).

REFERENCES

Allaert, F., Gentile, G., Baes, M., et al. 2015, A&A, 582, da Cunha, E., Charlot, S., Dunne, L., Smith, D., & A18 Rowlands, K. 2012, in IAU Symposium, Vol. 284, The Alton, P. B., Bianchi, S., Rand, R. J., et al. 1998, ApJL, Spectral Energy Distribution of Galaxies - SED 2011, ed. 507, L125 R. J. Tuffs & C. C. Popescu, 292–296 Alton, P. B., Xilouris, E. M., Bianchi, S., Davies, J., & Dalcanton, J. J., Yoachim, P., & Bernstein, R. A. 2004, Kylafis, N. 2000, A&A, 356, 795 ApJ, 608, 189 Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., Darg, D. W., Kaviraj, S., Lintott, C. J., et al. 2011, et al. 2013, A&A, 558, A33 MNRAS, 416, 1745 Baes, M., & Dejonghe, H. 2001a, MNRAS, 326, 722 —. 2010a, MNRAS, 401, 1043 —. 2001b, MNRAS, 326, 733 —. 2010b, MNRAS, 401, 1552 Baes, M., Verstappen, J., De Looze, I., et al. 2011, ApJS, De Geyter, G., Baes, M., Camps, P., et al. 2014, MNRAS, 196, 22 441, 869 Baldry, I. K., Liske, J., Brown, M. J. I., et al. 2018, De Geyter, G., Baes, M., Fritz, J., & Camps, P. 2013, MNRAS, 474, 3875 A&A, 550, A74 Bianchi, S. 2007, A&A, 471, 765 De Geyter, G., Baes, M., De Looze, I., et al. 2015, MNRAS, —. 2008, A&A, 490, 461 451, 1728 Bianchi, S., & Xilouris, E. M. 2011, A&A, 531, L11+ de Jong, J. T. A., Verdoes Kleijn, G. A., Kuijken, K. H., & Bridge, J. S., Hayes, M., Melinder, J., et al. 2018, ApJ, 852, Valentijn, E. A. 2013, Experimental Astronomy, 35, 25 9 de Jong, J. T. A., Verdoes Kleijn, G. A., Boxhoorn, D. R., Cameron, E. 2011, PASA, 28, 128 et al. 2015, A&A, 582, A62 Casteels, K. R. V., Bamford, S. P., Skibba, R. A., et al. de Jong, J. T. A., Kleijn, G. A. V., Erben, T., et al. 2017, 2013, MNRAS, 429, 1051 A&A, 604, A134 14 Holwerda et al. de Looze, I., Baes, M., Bendo, G. J., et al. 2012, MNRAS, Kelvin, L. S., Driver, S. P., Robotham, A. S. G., et al. 427, 2797 2014, MNRAS, 439, 1245 Dijkstra, M., Gronke, M., & Venkatesan, A. 2016, ApJ, Kruk, S. J., Lintott, C. J., Bamford, S. P., et al. 2018, 828, 71 MNRAS, 473, 4731 Dijkstra, M., & Wyithe, J. S. B. 2012, MNRAS, 419, 3181 Lintott, C., Schawinski, K., Bamford, S., et al. 2011, Dijkstra, M., Wyithe, S., Haiman, Z., Mesinger, A., & MNRAS, 410, 166 Pentericci, L. 2014, MNRAS, 440, 3309 Lintott, C. J., Schawinski, K., Slosar, A., et al. 2008a, Driver, S. P., Norberg, P., Baldry, I. K., et al. 2009, MNRAS, 389, 1179 Astronomy and Geophysics, 50, 050000 —. 2008b, MNRAS, 389, 1179 Driver, S. P., Hill, D. T., Kelvin, L. S., et al. 2011, Liske, J., Baldry, I. K., Driver, S. P., et al. 2015, MNRAS, MNRAS, 413, 971 452, 2087 Driver, S. P., Andrews, S. K., da Cunha, E., et al. 2018, Masters, K. L., Nichol, R. C., Hoyle, B., et al. 2011, MNRAS, 475, 2891 MNRAS, 411, 2026 Fortson, L., Masters, K., Nichol, R., et al. 2012, Galaxy Masters, K. L., Nichol, R. C., Haynes, M. P., et al. 2012, Zoo: Morphological Classification and Citizen Science, MNRAS, 424, 2180 ed. M. J. Way, J. D. Scargle, K. M. Ali, & A. N. Misiriotis, A., Popescu, C. C., Tuffs, R., & Kylafis, N. D. Srivastava, 213–236 2001, A&A, 372, 775 Grootes, M. W., Tuffs, R. J., Popescu, C. C., et al. 2013, Mosenkov, A. V., Allaert, F., Baes, M., et al. 2016, A&A, ApJ, 766, 59 592, A71 Hart, R. E., Bamford, S. P., Willett, K. W., et al. 2016, Obric, M., West, A. A., & Dalcanton, J. 2006, in American MNRAS, 461, 3663 Astronomical Society Meeting Abstracts, 97.08 Holwerda, B. W. 2005, PhD thesis, Proefschrift, Pastrav, B. A., Popescu, C. C., Tuffs, R. J., & Sansom, Rijksuniversiteit Groningen, 2005 A. E. 2012, in IAU Symposium, Vol. 284, The Spectral Holwerda, B. W., Bianchi, S., Baes, M., et al. 2011, “The Energy Distribution of Galaxies - SED 2011, ed. R. J. Spectral Energy Distribution of Galaxies” Proceedings Tuffs & C. C. Popescu, 306–308 IAU Symposium No. 284, 2012 R.J. Tuffs & C.C. Popescu, C. C., Misiriotis, A., Kylafis, N. D., Tuffs, R. J., Popescu, eds, 7, 128 & Fischera, J. 2000, A&A, 362, 138 Holwerda, B. W., Böker, T., Dalcanton, J. J., Keel, W. C., Popescu, C. C., Tuffs, R. J., Dopita, M. A., et al. 2011, & de Jong, R. S. 2013, MNRAS, 433, 47 A&A, 527, A109+ Holwerda, B. W., Dalcanton, J. J., Keel, W. C., Boeker, T., Popesso, P., Biviano, A., Rodighiero, G., et al. 2012, A&A, & de Jong, R. S. 2011, MNRAS 537, A58 Holwerda, B. W., Dalcanton, J. J., Radburn-Smith, D., Saftly, W., Baes, M., De Geyter, G., et al. 2015, A&A, 576, et al. 2012a, ApJ, 753, 25 A31 Holwerda, B. W., Bianchi, S., Böker, T., et al. 2012b, A&A, Schechtman-Rook, A., & Bershady, M. A. 2013, ApJ, 773, 541, L5 45 Hoyle, B., Masters, K. L., Nichol, R. C., et al. 2011, Schechtman-Rook, A., Bershady, M. A., & Wood, K. 2012, MNRAS, 415, 3627 ApJ, 746, 70 Hughes, T. M., Baes, M., Fritz, J., et al. 2014, A&A, 565, Shabala, S. S., Ting, Y.-S., Kaviraj, S., et al. 2012, A4 MNRAS, 423, 59 Hughes, T. M., Foyle, K., Schirm, M. R. P., et al. 2015, Stark, D. P., Walth, G., Charlot, S., et al. 2015, MNRAS, A&A, 575, A17 454, 1393 Hunter, J. D. 2007, Computing In Science & Engineering, Toomre, A. 1964, ApJ, 139, 1217 9, 90 Verstappen, J., Fritz, J., Baes, M., et al. 2013, A&A, 556, Jones, E., Oliphant, T., Peterson, P., & Others. 2001, A54 SciPy: Open source scientific tools for Python Wijesinghe, D. B., Hopkins, A. M., Sharp, R., et al. 2011a, Kaviraj, S., Ting, Y.-S., Bureau, M., et al. 2012a, MNRAS, MNRAS, 410, 2291 423, 49 Wijesinghe, D. B., da Cunha, E., Hopkins, A. M., et al. —. 2012b, MNRAS, 423, 49 2011b, MNRAS, 415, 1002 Keel, W. C., Manning, A. M., Holwerda, B. W., et al. 2013, Willett, K. W., Lintott, C. J., Bamford, S. P., et al. 2013, PASP, 125, 2 MNRAS, 435, 2835 Dust Lanes In Edge-On Galaxies in GAMA/KiDS 15

Wright, A. H., Driver, S. P., & Robotham, A. S. G. 2018, Wright, A. H., Robotham, A. S. G., Driver, S. P., et al. 2017, MNRAS, 470, 283 MNRAS, 480, 3491 Xilouris, E. M., Byun, Y. I., Kylaﬁs, N. D., Paleologou, E. V., & Papamastorakis, J. 1999, A&A, 344, 868