Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "Galaxy Zoo: dust lane early-type galaxies are tracers of recent, gas-rich minor mergers"

See other formats


Mon. Not. R. Astron. Soc. 000, 000-000 (0000) Printed 5 April 2012 (MN WI^ style file v2.2) 

Galaxy Zoo: Dust lane early-type galaxies are tracers 
-^. of recent, gas-rich minor mergers 

o 

^; Stanislav S. Shabala^'^*, Yuan Sen Ting^'^, Sugata Kaviraj^'^, Chris Lintott^, 

D- 
[-Ih ; R. Mark Crockett^, Joseph Silk^, Marc Sarzi^, Kevin Schawinski^, 

^ '■ Steven P. Bamford^ and Edd Edmondson^ 



o 



o 

m 
^^ 

o 



* Email: Stanislav.Shabala@utas.edu.au 



( | ■ ^ School of Mathematics & Physics, University of Tasmania, Private Bag 37, Hobart 7001, Australia 
I ' Oxford Astrophysics, Denys Wilkinson Building, Keble Road, Oxford OXl 3RH, UK 

o. 

r^ _ ^ Ecole Poly technique, 91128 Palaiseau Cedex, France 

C^ . * Blackett Laboratory, Imperial College London, London SW7 2AZ, UK 

^ Centre for Astrophysics Research, University of Hertfordshire, College Lane, Hatfield ALIO 9AB, UK 

(N. 

^ ' ® Yale Center for Astronomy and Astrophysics, Yale University, P.O. Box 208121, New Haven CT 06520, USA 



^ Centre for Astronomy & Particle Theory, University of Nottingham, University Park, Nottingham NG7 2RD, UK 
* Institute of Cosmology and Gravitation, University of Portsmouth, Burnaby Road, Portsmouth POl 3FX, UK 

t This publication has been made possible by the participation of more than 250 000 volunteers in the Calaxy Zoo 2 project. 
Their contributions are individually acknowledged at \ http://zoo2. galaxyzoo.org/authors \ 



X: 

;_| _ 5 April 2012 



ABSTRACT 

We present the second of two papers concerning the origin and evolution 
of local early-type galaxies exhibiting dust features. We use optical and radio 
data to examine the nature of AGN activity in these objects, and compare 
these with a carefully constructed control sample. 

We find that dust lane carly-typc galaxies are much more likely to host 
emission-line AGN than the control sample galaxies. Moreover, there is a 
strong correlation between radio and emission-line AGN activity in dust lane 
early-types, but not the control sample. Dust lane early-type galaxies show 
the same distribution of AGN properties in rich and poor environments, sug- 



Shahala et al. 



gesting a similar triggering mechanism. By contrast, this is not the case for 
early- types with no dust features. These findings strongly suggest that dust 
lane early- type galaxies are starburst systems formed in gas-rich mergers. 
Further evidence in support of this scenario is provided by enhanced star for- 
mation and black hole accretion rates in these objects. Dust lane early-types 
therefore represent an evolutionary stage between starbursting and quiescent 
galaxies. In these objects, the AGN has already been triggered but has not as 
yet completely destroyed the gas reservoir required for star formation. 

Key v^rords: galaxies: evolution — galaxies: interactions — galaxies: active 
— galaxies: elliptical and lenticular 



1 INTRODUCTION 

Dust lane early-type galaxies are those galaxies classified in Galaxy Zoo '42 as early-type 
galaxies with dust lane features. The classification procedure and construction of the dust 
lane sample are described in detail in a companion paper, Kaviraj et al. (2012; hereafter 
Paper 1). In that paper, we showed that dust lane early-type galaxies often exhibit morpho- 
logical disturbances, show signs of recent star formation, and have stellar ages that are older 
than starburst early-types, but younger than the overall early-type population. We therefore 
argued that these objects are starburst systems, and suggested that they might arise from 
gas-rich minor mergers. While IGM accretion is also a plausible mechanism, we ruled it out 
in this case due to the presence of large amounts of dust, and the disturbed nature of most 
galaxies. 

In the present paper we test this paradigm by examining the Active Galactic Nucleus 
(AGN) properties of dust lane early-type galaxies. Numerous theoretical and observational 
arguments connect star formation and AGN triggering in galaxies. The basic point is that 
similar conditions (namely, the presence of cold gas) are required for both a nuclear starburst 
and AGN fuelling. Once the supermassive black holes that reside at galaxy centres become 
active, radiative and kinetic feedback from these objects can profoundly affect their environ- 
ments through heating and expulsion of gas. AGN feedback has been invoked to explain the 



cooling flow problem (Binney &: Tabor 1995), and accounts for both the shape of bright end 



of the optical luminosity function (IShabala Sz Alexander 20091) and the strong correlation 



^ http://zoo2.galaxyzoo.org 



Dust lane galaxies as tracers of gas-rich mergers 3 
between black hole and bulge mass (Magorrian et al. 1998 Haring & Rix 2004) in the local 
Universe. 

A number of diagnostics are used to identify AGN. Perhaps the best known is the so- 
called BPT fIBaldwin et al. 198ip analysis, which relies on emission-line measurements to 
separate star-forming galaxies from AGN. Synchrotron emission from AGN jets and jet- 
inflated cocoons manifests itself at radio frequencies. X-ray and infra-red diagnostics are 
also used. 

Previous studies have shown that Seyfert and LINER elliptical hosts are more likely to 
contain dust than inactive ellipticals (Simoes Lopes et al. 2007 IMartini et al. 2003]) . How- 
ever, it is unclear whether this finding can be extended to other AGN classifications. Re- 
cently, Best et al. ( ]2005bll showed that radio and emission-line AGN activity are largely 
independent at low radio luminosities, in contrast to the well-known correlation at high 
luminosities ( Rawlings fc Saunders 199T ). This led Hardcastle et al. ( I2007P to suggest an 
environmental origin for this dichotomy. Bright radio sources preferentially live in the field; 
these are also the objects that often exhibit emission-line AGN properties. By contrast, 
low-luminosity radio AGN tend to live in denser environments, and show little emission-line 
activity. Hardcastle et al. argued that fundamentally different AGN fuelling mechanisms are 
likely to be at work in these two scenarios, with radio-loud objects in dense environments 
mainly fuelled by the relatively slow bremsstrahlung cooling out of the hot X-ray gas halo; 
while in poor environments mergers are primarily responsible for the sudden increase in cold 
gas content. The properties of the black hole accretion disk are different in the two cases. At 
high accretion rates associated with mergers, the disk assumes the standard geometrically 
thin, optically thick shape (e.g. Shakura & Sunyaev 1973); this disk is radiatively efficient, 
and can also produce a radio jet. Thus, both emission-line and radio AGN are detected. At 
low accretion rates, however, the accretion disk puffs up and becomes radiatively inefficient, 
yielding a powerful radio jet but no corresponding emission lines (IMeier 200ip . Therefore, 
a detailed analysis of both optical and radio AGN properties of our sample can help recon- 
struct the origins of the gas used in both star formation and AGN fuelling, and indicate 
whether dust lane early-type galaxies are indeed produced in galaxy interactions. Moreover, 
sizes and luminosities of radio AGN can be used to constrain their ages (IShabala et al. 2008P 
and therefore place these objects on an evolutionary sequence. 

The homogeneity and size of our sample of dust lane early-type galaxies allows us to study 
the AGN triggering mechanisms via a combination of photometry, spectroscopy and radio 



Shahala et al. 







r 






S 60 






-\j 




- 


o 40 






1 




fe 






^^ 1 1 1 


^ ?0 




^ 




h ■ 








1 





) 


S 60 




r^ 


- 


g, 








o 40 






- 


b 


1^ 




2 ?0 


- ^ \ 




f ^ ^ ^ 



0,00 0.02 0.04 



0.08 O.'O 



9.5 10.0 10.5 n 




-23 -22 



(a) 



(b) 



(c) 



Figure 1. Rcdshift {top), stellar mass {middle) and r-band absolute magnitude {bottom) distribution of the dust lane early-type 
galaxy sample. The dashed lines show the adopted redshift cut. 

data. We create well-defined control samples, facilitating the construction of an evolutionary 
sequence and uncovering the significance of the observed dust deatures. 

The paper is structured as follows. In Section [2] we present our sample of dust lane 
early-type galaxies as well as a relevant control sample. In particular, we mimic the salient 
properties of dust lane galaxies in constructing the control sample. Section |3] outlines the 
AGN selection procedure using two alternative definitions: radio luminosity and emission 
lines. We present our results in Section HI Environmental dependence of dust lane AGN 
properties indicates that the origin of these objects is merger- driven. We show that these 
mergers are necessarily recent, and gas-rich. Finally, we conclude in Section [5l 

Throughout the paper, we adopt the Hubble constant of Hq = 72kms~^Mpc~^. 



2 SAMPLE CONSTRUCTION 

2.1 Dust lane early-type galaxies 

The selection of early-type galaxies and a subsample of these exhibiting dust lane features 
from SDSS DR7 is described in detail in Paper 1. The only slight difference between the 
samples analyzed here and in Paper 1 is the inclusion of bulge-dominated Sa galaxies in the 
present work, which were added to improve statistics. Sa galaxies constitute 23 percent of the 
dust lane sample, and exhibit very similar redshift distribution and disturbed fraction values 
to the rest of the sample. We made sure that their inclusion did not bias our conclusions by 
repeating the analysis without these objects, obtaining the same results. Figure [1] shows the 
stellar mass and redshift distribution of dust lane early-type galaxies. 

Our control sample consists of all early-type galaxies in the redshift range 0.01 < z < 
0.07. The upper limit on redshift corresponds to the maximum distance at which dust 
lanes can be reliably detected: as can be seen from Figure [H at redshifts z > 0.08 the 
fraction of detected dust lane features begins to drop, presumably due to the difficulty 



Dust lane galaxies as tracers of gas-rich mergers 5 

of identifying these with a naked eye. The low-redshift cutoff is imposed for two reasons. 
Firstly, the radio-optical matching performed in the following section becomes increasingly 
inaccurate (as the same physical separation corresponds to a larger angular separation at 
lower redshift). Secondly, V^/Knax corrections for Malmquist bias (e.g. Shabala et al. 2008) 
to the radio luminosity function (RLF) become extreme if many low-luminosity (and hence 
low redshift) objects are included. 

In principle, the variation in redshift and especially mass distribution between the dust 
lane and control samples can bias our conclusions. Therefore, for the control sample we 
mimic the redshift, stellar mass and absolute magnitude profiles of the relevant dust lane 
subsample. 



2.2 Radio-optical matching 

We follow the procedure outlined in Shabala et al. (I2008P and Best et al. (I2005al) to construct 
radio-optical catalogues for dust lane early-type galaxies and the relevant control samples. Of 
the two large-area surveys at 1.4 GHz, Faint Images of the Radio Sky at Twenty Centimetres 
(FIRST; Becker et al. 1995) provides much better positional accuracy than the NRAO VLA 
Sky Survey (NVSS; Condon et al. 1998). We paired sources from the FIRST catalogue with 
SDSS galaxies using a number of separation distances ranging from 1-30 arcsec. The point 
at which the number of matches stays approximately unchanged for any further increase 
in separation distance provides a good estimate of the best pairing distance estimate. This 
method yielded a best separation distance of a few arcseconds. 

A different estimate is provided by generating a mock catalogue, constructed by offset- 
ting all sources in the SDSS catalogue by 5 arcmin in right ascension. Both the mock and the 
actual SDSS catalogues were then paired with FIRST for a range of separation distances. 
At the largest separation distance, we expect all real matches to be captured, plus an un- 
known number of false matches (due to the allowed separation distance being too large). 
An approximate number of false matches was obtained by subtracting the number of total 
matches for the real catalogue from the corresponding value for the mock catalogue. 

The best matching radius is obtained by maximizing the completeness, and minimizing 
contamination of the sample. Here, completeness is defined as the fraction of genuine optical- 
radio pairs being identified; and contamination as the number of false positives. Table [1] led 
us to adopt a separation distance of 3 arcsec. 



6 Shahala et al. 



Separation 


Real matches 


Completeness 


False matches 


Contamination 


(arcsec) 




(per cent) 




(per cent) 


0.5 


1211 


59.7 


1 





1 


1742 


85.9 


1 





2 


1948 


95.9 


6 


0.3 


3 


1974 


96.8 


12 


0.6 


5 


2002 


97.7 


23 


1.1 


15 


2174 


98.9 


170 


8.4 


30 


2625 


100 


599 


29.6 



Table 1. Completeness and contamination as a function of SDSS-FIRST matching radius. The number of real matches is 
approximated by subtracting mock catalogue counts from the actual catalogue. A matching radius of 3 arcsec is adopted. 

Best et al. (l2UU5ap found that many FIRST sources with offsets between 3 and 10 arcsec 
were found to be extended sources oriented close to the direction of the offset between the 
optical and radio positions. We followed these authors in refining the selection procedure 
to include FIRST sources which are offset from the SDSS position by more than 3 arcsec 
but less than 10 arcsec, and are oriented within 30° of the vector defined by the SDSS and 
FIRST coordinates. This resulted in an inclusion of an extra 20 sources (1% of the sample) 
in the real catalogue, but only 2 (0.1%) in the mock one, improving completeness without 
contaminating the sample. 

We note that in the following analysis, NVSS maps (which are more sensitive to large- 
scale structure, but have less accurate position identifications) are used in parallel with 
FIRST for the galaxies identified as having a radio counterpart as outlined above. This was 
done to prevent any bias in observed radio AGN sizes and ages that would result from only 
using one of these surveys. 

3 AGN SELECTION 

3.1 Infra-red - radio correlation 

Most galaxies hosting radio AGN will also contain a contribution to radio luminosity from 
synchrotron emission associated with supernova-driven shocks. This needs to be subtracted 
from the total luminosity if the AGN component is to be defined properly. Using the relation 
of Yun et al. (I200ip . at 1.4 GHz the radio luminosity due to star formation is 



i;^:^ = 1.04 X 10^2 r "^ 



WHz~i \Moyr-i; ^ ' 

Perhaps the best estimate of the star formation rate comes from far infra-red data, 
where it is unobscured. However, as Figure |2]a illustrates, selection of early-type galax- 
ies with IRAS counterparts is biased towards blue early-type galaxies. A comparison of 



Dust lane galaxies as tracers of gas-rich mergers 7 



3.5 1 ' ' ; 




10.0 10.2 10.4 10.6 10.8 11.0 
log M g[^^3 / Mq 

(a) 




-2 2 

log SFR soss / Moyr"' 

(b) 



Figure 2. Top: Colours of early-type galaxies with an IRAS counterpart (points), compared with all early-type galaxies 
(contours). IRAS selection is biased towards lower masses and bluer colours. Bottom: Comparison of star formation rates 
derived from IRAS and SDSS. Contours and points represent the same data set of all SDSS galaxies with IRAS identifications. 
A good correlation is seen, with SDSS typically slightly underestimating the star formation rates. Solid line gives the median 
value, and dashed lines the 1.5a- upper and lower bounds. In both plots the 10, 30, 50, 70, 90 % contours are shown. 



IRAS flKennicutt 1998P and SDSS-derived f lBrinchmann et al. 20041 ITremonti et al. 2004p 
star formation rates for all galaxies in the redshift range of interest in Figure |2]6 shows a 
good correlation between the two. The lower star formation rates in SDSS arise due to a 
combination of extinction and smaller aperture sizes. We therefore use the best-fit relation to 
correct for the extinction associated with dust-corrected SDSS-derived star formation rates. 
Excess radio luminosities, which we attribute to AGN activity, are calculated by subtracting 
a star-formation contribution from the total luminosity. 

There is inherent scatter in the SFR-Li 4 sf relation given by Equation [H which can 
result in misclassification of star-forming galaxies as radio AGN. We therefore adopted a 
conservative requirement that, for an object to be classified as a radio AGN, the total radio 
luminosity must significantly exceed the mean value expected due to star formation alone. 
A number of cuts in radio luminosity were examined before adopting a 1.5cr cut (i.e. Li 4 
must exceed Li.4,sf by more than 1.5 times the scatter in the IRAS-SDSS SFR relation). 
Taking a la cut results in classification of many blue early-type galaxies (as given by their 
(m — r) colours) as radio- loud AGN. A 2cr cut yields the same results as our adopted cut, 
but with poorer statistics. 

It is worth pointing out that here we are implicitly using nuclear star formation rates. 
As dust lane galaxies may exhibit more extended star formation than the control sample 
galaxies, considering only nuclear fluxes could in principle make dust lane hosts more likely 
to be classified as Seyferts or LINERs in the emission-line diagnostic in Section 13.21 How- 
ever, the dominant uncertainty of any such classification is much more likely to lie in the 
AGN/star-forming separation in the BPT plot rather than aperture effects. On the other 



8 Shabala et al. 

hand, using nuclear fluxes ensures that sources identified as LINERs are genuine AGN rather 
than distributed old stellar populations. 

This procedure identifies 65 radio loud AGN (out of 484; 13%) in the dust lane sample, 
and 1188 out of 32 590 (3.6%) in the control sample of early-type galaxies. 

3.2 Emission line AGN 

An alternative way to identify AGN hosts is through the use of the emission-line diagnostic 
(the so-called BPT diagram; Baldwin et al. 1981). We refer the reader to Paper 1 for details 
of this process. Essentially, the 484 dust lane early-type galaxies, and 32 590 control sample 
early-types were classified according to their O [III]/H/3 and N[II]/Hq; line ratios. Star- 
forming galaxies occupy a locus in the O [III]/H/3 - N [II] /Ha plane. Objects showing excess 
O [III] or N [II] emission, but located close to the star-forming locus are classified as transition 
objects, i.e. galaxies which host both significant levels of star formation and AGN activity. 
Galaxies exhibiting even stronger O [III] or N [II] lines are classified as Seyfert AGN or Low- 
lonization Emission Line Regions (LINERs). This analysis was performed for both the dust 
lane and control samples. 

4 AGN TRIGGERING AND THE AGN-STARBURST CONNECTION 

4.1 Environments 

As outlined in Section [H a major motivation for considering two different modes of AGN 
activity is to probe both the AGN triggering mechanisms, and dust lane galaxy formation 
scenarios. Radio and emission-line AGN activity coincide for high-power radio sources, while 
low-luminosity radio AGN are identified with little optical activity. Hardcastle et al. (I2007p 
suggested that such a scenario is consistent with high excitation line AGN being triggered 
by gas-rich merger^l, where large amounts of gas are added to the galaxy in a relatively 
short period of time; while radio-only AGN can be fuelled by cooling out of the hot halo gas 
(Shabala et al. 2008; Best et al. 2005b). In this scenario, a strong environmental dependence 
is expected, with merger-driven AGN being found in the field, and radio-only AGN residing 
in overdensities such as groups and clusters. 

^ By 'gas-rich' we don't necessarily mean a merger between two spiral galaxies. This is simply any merger which can supply a 
substantial amount of gas to the merger remnant over a short period of time. Cannibalism of dwarf galaxies and minor mergers 
with high merger ratios (> 1 : 10) could plausibly meet this condition. 



Dust lane galaxies as tracers of gas-rich mergers 9 




(a) 







12 13 14 15 



(b) 



Figure 3. Top: Distribution in environmental parameter p as defined by Schawinski et al. I I2007I I for dust lane early-type galaxies 
(dashed) and a matched sample of all early-type galaxies (solid line) with 0.01 < z < 0.07. Dust lane early-type galaxies reside 
in less dense environments than "average" early-type galaxies. Bottom: Relationship between p and the statistical density 
estimator of Yang et al. II2007II . Vertical "error bars" correspond to 25 and 75 percentile values. 



Figure [3]a shows the distribution of environments for dust lane early-type galaxies and 
the matched control sample. In Paper 1, environment information was obtained by cross- 
matching our samples with the Yang et al. (l2UU7p group catalogue. This statistical approach 
works well for large samples. In this paper, we quantify the nearest neighbour density within 
2 Mpc using the environment parameter p of Schawinski et al. ( I2007p . Figure |3]& shows the 
strong correlation between the two quantities. In what follows, we adopt p as a density 
indicator; however, all our findings are recovered when halo mass estimates of Yang et al. 
are used instead. Dust lane early-types are preferentially located in poor environments. The 
difference between the distributions in Figure |3]a, as given by a Kolmogorov-Smirnov test, 
is significant at the 99 percent level. 

It is worth noting that were are confident of having removed any potential mass depen- 
dence in the above environment result; nor is it subject to a selection bias. In all results 
described here, we made sure that for all control samples we mimicked the dust lane early- 
type mass, redshift and brightness distributions, as shown in Figure [H Given the strong 
evolution of many galaxy properties with mass (e.g. Kauffmann et al. 2003; Shabala et al. 
2008), this is crucial to ensuring that our findings are not simply a selection effect. Such 
careful construction of control samples is even more important in light of the visual selection 
employed in extracting dust lane early-type galaxies. 



4.2 Radio loud fractions 

We now consider the radio AGN properties of early-type galaxies. Figure H] shows the cu- 
mulative fraction of galaxies hosting a radio AGN brighter than a given luminosity. The 



10 Shabala et al. 









9.8 < log M „„, 


, < 10.5 








log M „j. 


, > 10.5 ■ 


^ 0.1000 
s 
! 

-^ 0.0100 
A 


^ 




" — ^.^ " 


-, 


^ 0.1000 
s 

^ 0.0100 
A 


r * 


\ 


^^^^ 


c 
o 












c 

o 












D 0.0010 


- 




All ETGs 




~ 


o 0.0010 


r 




- All ETGs 




" 
















• 


Dust lane ETGs 










• 


Dust lane ETGs 






0.0001 


, 




III 






0.0001 






1 1 1 1 


, 





21.6 21.8 22.0 22.2 22.4 22.6 22.8 23.0 
log L ,... „„„ / W Hz "' 



21.6 21.8 22.0 22.2 22.4 22.6 22.8 23.0 
log L ,4 „„„ / W Hz "' 



(a) Low-mass galaxies 



(b) High-mass galaxies 



Figure 4. Top: Cumulative fractional radio luminosity functions for low-mass dust lane (points) and a matched sample of 
low-mass early-type galaxies (shaded region) with 0.01 < z < 0.05. Standard V/Vmax corrections are applied. The dashed line 
corresponds to the completeness limit at the back of the volume. Bottom: Same as the left panel but for massive galaxies. 
Massive dust lane early-types host significantly fewer radio AGN than the control sample. 



fractions were obtained by applying standard V^/Knax corrections to radio source counts in 
order to correct for Malmquist bias. We adopted the redshift range 0.01 < z < 0.05 for 
our samples in order for these corrections to not dominate genuine source counts. Since the 
colours and ages of local galaxies show a fairly well-defined bimodality, separating into the 
red sequence and blue cloud (with relatively few green valley objects) at a stellar mass of 
~ 3 X lO^°M0 (lKauffmann et al. 2003cp . we split our samples into high and low stellar mass 
bins at this value. Stellar masses, derived in Paper 1, are based on the calibrations of Bell 
et al. (120031) . 

At low masses, dust lane early-type galaxies show a marginally higher rate of AGN activ- 
ity than the matched control sample of early-type galaxies, especially at lower luminosities. 
In massive galaxies, this result is reversed. One must be careful about the interpretation 
of these findings. In Figure O a similar result is obtained when the samples are split up 
by environment, rather than mass. Taking a cut at p = 0.05, corresponding roughly to 
the transition between field and group/cluster environments (Schawinski et al. 2007; see 
also Figure [3f>), dust lane and control sample early-type galaxies show comparable rates of 
AGN activity; while in groups/clusters the control sample early-types are more likely to 
host AGNs. Since massive early-type galaxies are preferentially found in clusters, the result 
of Figure mis in fact an environment, rather than a mass, dependence. These findings are 
confirmed quantitatively by chi-squared tests at 99 percent level. 

Interpretation of these results becomes easier when dust lane and the all early-types 
control sample are considered separately. As Figure [6] shows, the radio AGN fraction in 
dust lane early-type galaxies is independent of environment. Moreover, this fraction is also 



Dust lane galaxies as tracers of gas-rich mergers 11 





' 


1 1 1 


' ' 


' 


- 










Field 


- 


0.1000 


" ^ 


~*~— i~_ 






- 


0,0100 


^ 


^^ 


"^ 


^^^^ 




-_ 








■ t 


'. 


0.0010 


- 










- 










m 


Dust lone ETGs 








- 


0.0001 


, 1 , , . 


1 , . , 1 . , , 1 , , . 


1 , . . 


... 1 













CI 


ster 


0.1000 


: i 


"^^^^^7^ 


--^ 


": 


J 0.0100 

A 


r 






~: 


c 

■^ 0.0010 


r 








■ 










m 


Dust lane ETGs 






- 


0.0001 


. 1 , . , 


1 . , , 1 , , . 1 , . , 1 . , , 1 









21.6 21.8 22.0 22.2 22.4 22.6 22.8 23.0 

log L 1.4, excess / W Hz "' 

(a) Field 



21.6 21.8 22.0 22.2 22.4 22.6 22.8 23.0 
l°9 L ,.4, e,„ss / W Hz "' 



(b) Group/Cluster 



Figure 5. Same as Figure l4l but for environments. Top: p < 0.05. Bottom: p > 0.05. Dust lane early-types have lower AGN 
fractions than the control sample in clusters, while there is no difference in the field. 




22.0 22.2 22.4 
log L , ,^ „„„ / W Hz 




22.0 22.2 22.4 22.6 22.J 
log L ,4, „„„ / W Hz '' 



(a) Dust lane early-typos 



(b) Control sample 



Figure 6. Same as Figure [5] but separated by environment for each of the two samples. Top: dust lane early- type galaxies. 
Bottom: control sample. In both cases group/cluster galaxies are shown by the dark shaded region, and field galaxies as a 
light shaded region. There is no difference in radio AGN fractions between dust lane hosts in clusters and the field, while a 
significant enhancement in the cluster AGN fraction is seen in the control sample. This enhancement comes about due to a 
"cooling mode" of AGN fuelling available to massive early-type galaxies in clusters. In the field, the AGN fraction is set by 
the merger rate. The lack of AGN fraction dependence on environment for dust lane hosts strongly suggests that these objects 
trace gas-rich mergers. 



consistent with the radio AGN fraction of all early-type galaxies found in poor environments. 
In clusters this fraction rises significantly for early-type galaxies. Again, our conclusions are 
confirmed by chi-squared tests at the 99 percent level. 

This latter result suggests the following interpretation. Radio AGN can be fuelled either 
by gas cooling out of the hot halo, or a sudden supply of cold gas through a gas-rich merger 
event. The former scenario is more likely in massive early- type galaxies at the centres of 
clusters. Field/low-mass early-types are less likely to be fuelled by this mechanism, with 
gas-rich mergers feeding the black hole in AGN hosts. Since in the local Universe such gas- 
rich merger events are much more rare than cooling episodes, cluster/high-mass early-type 
galaxies are more likely to host radio AGN than their field/low-mass counterparts. 

The dust lane early-type population, both in the field and in clusters, shows the same 



12 Shabala et al. 



log (M/Mq) < 10.5 log (M/Mq) > 10.5 Total 



Dust lane early-types 0.83 (15/18) 0.92 (43/47) 0.89 (58/65) 

Control early-types 0.53 (18/34) 0.25 (61/240) 0.29 (79/274) 

Table 2. Probability of finding an emission-line AGN given a radio-loud AGN. Entries denote the probability for the given 
galaxy type, while source counts are given in brackets. Dust lane early-type galaxies exhibit a strong correlation between 
emission-line and radio AGN activity. 

distribution of radio properties as the field control population. Therefore, if mergers are the 
major trigger of radio AGN activity in all field early-types, this is also the case for dust lane 
early-type galaxies, regardless of whether they are located in rich or poor environments. In 
other words, it appears as though a dust lane feature serves as a proxy for a recent gas-rich 
merger. This paradigm is supported by the finding in Paper 1 that almost all dust lane 
galaxies exhibit morphological disturbances and recent starbursts. 

According to this scenario, all dust lane early-types should have undergone a recent 
gas-rich merger (and hence presumably host a detectable AGN), while only some of the 
early-types in the control sample would have done the same. Figures [5] and [6l however, show 
that the radio-loud fractions for these objects are similar. This is most likely due to our 
conservative {1.5a above emission due to star formation) radio luminosity cut adopted for 
AGN identification. 

4.3 AGN triggering 

One way the above paradigm can be tested is by comparing radio and emission-line AGN 
activity in dust lane early-type galaxies and the control sample. If gas-rich mergers are 
indeed responsible for fuelling the AGN in dust lane early-types, one would expect a much 
higher fraction of radio AGN in these objects to also exhibit emission-line AGN activity, 
compared to objects in which gas cooling is the major AGN trigger. In Table [2] we present 
the fraction of radio AGN hosts that are also identified as emission-line AGN. Almost all 
objects identified as radio AGN also show up as emission-line AGN in dust lane galaxies, 
but this is far from the case in the control sample, consistent with findings of the previous 
section. 

Table [3] shows the fraction of various AGN types, as defined in Paper 1, in dust lane 
early-type galaxies and the control sample. The control sample shows a much lower level of 
emission-line AGN activity (< 20%) than dust lane early-types (> 70%). Taken together 
with the above result that only dust lane galaxies (but not the control sample) show a 
significant correlation between emission-line and radio AGN activity, our interpretation is 



Dust lane galaxies as tracers of gas-rich mergers 13 



Classification 



Control 



Dust lane early-types 



Weak 


0.78 (4510) 


0.12 (30) 


Star-forming 


0.06 (368) 


0.13 (32) 


Transition 


0.06 (343) 


0.28 (68) 


LINER 


0.06 (369) 


0.26 (62) 


Seyfert 


0.03 (160) 


0.21 (50) 



Emission-line AGN 0.15 (874) 
Total 1.0 (5750) 



0.74 (180) 
1.0 (242) 



Table 3. Emission-line AGN fractions by type for dust lane early-type galaxies and a matched control sample of all early-types 
with log (Mt/MQ) < 10.5. Emission-line AGN in the second-last row are defined as the galaxies classified as either transition 
objects, LINERs or Seyferts. Entries denote fractions of total population for the given galaxy type, while source counts are 
given in brackets. Note that the counts can be non-integral in the case of the matched control sample. Dust lane early-types 
exhibit a much higher level of emission-line AGN activity than the control sample. 



60 F 



E 20 



All ETGs 
Dust lone ETGs 




og tj / Gyrs 

Figure 7. Distribution of starburst ages for dust lane early-type galaxies and a matched control sample of all early-types. 
Median ages are shown by dashed lines. Dust lane early-type galaxies host starbursts that are younger than those in all 
early- type galaxies. 

that dust lane early-type galaxies correspond to a phase in which radio and optical AGN 
activity are triggered. 

4.4 Stellar ages 

If dust lane early-types are indeed associated with gas-rich mergers, an obvious question 
concerns their place in the evolution of a post-merger early-type galaxy. Following Kaviraj 
( I2010bp we estimate the age of the last starburst using the double colour {NUV — u) — {g — z), 
which is relatively dust insensitive. Figure [7] shows the starburst age distributions for the 
dust lane sample and a matched control sample of all early-type galaxies. The starburst ages 
of dust lane early-type galaxies are younger than the average early-type galaxy. 



4.5 Radio source ages 

As shown in the previous section, dust lane early-type galaxies have stellar populations 
that are younger than the overall early-type population. Thus, these objects are likely to be 
starburst or post-starburst galaxies, observed during the transition between the blue and 



14 Shabala et al. 

red- and- dead phases. The correlation between optical and radio AGN activity suggests gas- 
rich minor mergers (since major mergers are rare in the low-redshift Universe) may drive 
star formation and AGN activity. In this scenario, the central engine of the AGN is endowed 
with so much fuel that the accretion disk settles into a radiatively efficient, optically thick, 
geometrically thin state (e.g. Meier 2001) and yields both an observable radio jet and narrow- 
line emission. Tables [2] and |3] suggest this is not the case for the later (control early-types) 
stages of evolution. This is likely to be because the merger-driven AGN activity has mostly 
expelled the available gas, and the galaxy migrated to a quiescent state. The mean starburst 
age for the control sample is 700 Myrs, which is much larger than both the transition time 
of ~ 200 Myrs between the starburst and transition (i.e. starburst -|- AGN) phase argued by 
Schawinski et al. (2007; see also Figure [8]), and the maximum AGN lifetime of up to a few 
hundred Myrs (IShabala et al. 2008p . Therefore, most of these objects are not expected to 
host AGN that would have a common origin with the starburst. Conversely, with their mean 
starburst ages of ~ 300 Myrs, dust lane early-types fall in the "sweet spot" of being old 
enough to host an AGN, but not so old that AGN activity has been terminated. 

We test this paradigm by deriving radio AGN ages for those galaxies that were classified 
as having excess 1.4 GHz luminosity. This was done by generating a library of tracks in 
radio luminosity - linear size space. As discussed at length in Shabala et al. (I2008p . given 
a density profile for the atmosphere into which the radio source is expanding, and aspect 
ratio of the source, its luminosity and size allow jet power and age of the radio source to 
be determined uniquely. We use observations of local early-type galaxies f lAUen et al. 2006|) 
for the X-ray gas density profile, and adopt an axial ratio of Rt = 2. A number of caveats 
are associated with this analysis. Our assumption for the gas density profile may not be 
applicable for dust lane early-type galaxies, most of which show disturbed morphologies 



( Paper 1 ). Furthermore, hotspot contribution to radio luminosity can be comparable with 
that of the lobes at 1.4 GHz. However, Shabala et al. (I2008P estimate that the derived 
timescales are accurate to better than a factor of two. 

Importantly, the radio source models of Kaiser & Alexander (I1997P and Kaiser et al. 
(I1997P are only applicable for edge-brightened (Fanaroff-Riley type II) sources. On the other 
hand, most radio AGN in the local universe are core-dominated FR-I objects. However, most 
of the resolved objects in our sample are FR-IIs, for which the modelling is applicable. Only 
upper limits on ages can be placed for unresolved radio sources. 

Table |4] gives the size distribution of radio AGN for the dust lanes and the matched 



Dust lane galaxies as tracers of gas-rich mergers 15 



size < 30 kpc 



size > 30 kpc 



Total 



Dust lane early-types 0.16 (43/270) 0.008 (3/270) 0.17 (46/270) 

Control early-types 0.058 (481/8951) 0.0007 (36/8951) 0.058 (517/8951) 

Table 4. Fractional distribution of radio source sizes. There is no statistically significant difli'erence between dust lane early-type 
galaxies and the control sample. Most radio sources in the dust lane sample are unresolved with FIRST, and only upper limits 
on their sizes are available. 




200 400 600 

t 2 / Myr 

(a) starburst ages 



800 




800 



(b) starburst and radio ages 



Figure 8. Top: Distribution of starburst ages for various emission line classifications of dust lane early-type galaxies. Both 
radio-loud and radio-quiet objects are included. An evolutionary sequence of starburst — >■ transition object — > Seyfert/LINER 
is seen. Bottom: Comparison of starburst and radio source ages for dust lane early-types. There is no difference in starburst 
age distributions between radio-loud and radio-quiet AGN. 



control sample. The distributions are statistically indistinguishable at the 10 percent level. 
However, the number of dust lane radio AGN in our sample (46) is small, and most of these 
are unresolved. High resolution imaging will be required to address this issue properly. 

Figure [8] compares the derived radio source ages with the ages of stellar populations 
for dust lane early-type galaxies. Seyferts and LINERs have similar age distributions, with 
median values around 300 Myrs. There is no difference between radio-loud and radio-quiet 
emission-line AGN. We interpret this as evidence for the simultaneous triggering of the radio 
jet and line emission. Seyfert-like line emission will come from the radiatively efficient AGN 
disk, while LINER-like emission can come from either the disk or be distributed throughout 
the radio cocoon (e.g. Krause & Gaibler 2009). This picture is further supported by the 
100 — 150 Myr offset between radio and starburst ages in Figure [H]^ being consistent with 
the age difference between starburst and transition objects in Figure [H]a. 

Our interpretation of the origin and evolution of dust lane early-type galaxies is then 
as follows. A quiescent early-type galaxy undergoes a gas-rich minor merger. The resulting 
gas inflow triggers a starburst. Some time later (typically < 200 Myrs) the AGN switches 
on. Circumnuclear starbursts could be the reason for this delay, with powerful winds driven 
by OB stars and supernovae suppressing AGN fuelling temporarily (IDavies et al. 2007|l . 



16 Shabala et al. 

Alternatively, mass loss from newly formed stars could help fuel AGN activity, providing a 
possible mechanism of funneling gas towards the central engine. Once the AGN switches on, 
the accretion rate onto the central black hole is high enough (typically at least a few percent 
of the Eddington value) for the accretion disk to be in a classical, radiatively efficient thin 
disk state, and the AGN is observed in both emission lines and at radio wavelengths. This 
phase is accompanied by the presence of significant amounts of dust, and the early- type 
galaxy is observed as having a dust lane feature. Eventually, AGN feedback heats up and/or 
expells the gas from the galaxy (this is evidenced for example by the depletion of molecular 
gas reservoir on timescales of around 200 Myrs; Schawinski et al. 2009a). Both the star 
formation and AGN activity are truncated, with the galaxy returning to a quiescent state. 



4.6 Origin of dust 

In Paper 1 we showed that the mass of dust in the observed features is too great to come 
from stellar mass loss, suggesting strongly an external origin for this dust. Since dust and 
gas are typically coupled, one might expect dust lane early-type galaxies to be gas rich, 
consistent with the gas-rich merger origin of these objects. As shown in Paper 1, dust lane 
early-types do show enhanced levels of star formation compared to a control sample of all 
early-type galaxies. However, Figure [7] shows that galaxies with dust features also have 
younger starburst ages, and it is therefore possible that the higher star formation rates in 
these objects could simply be due to them having had less time to deplete their newly- 
acquired gas reservoirs. 

We alleviate this potential problem by requiring the dust lane control and sample galax- 
ies to span the same range in starburst age, 100 < ^2 < 400 Myrs. As Figure [8]a shows, this 
preferentially picks out the AGN population (as defined in the BPT plot), and should be rep- 
resentative of recent merging events in both the dust lane and overall early-type populations. 
We note that the stellar mass distributions of the two subsamples are very similar. 

Figure [9]a shows that, even after correcting for the difference in starburst age distribu- 
tions, dust lane galaxies exhibit star formation rates that are higher by a factor of three 
than the control sample. If dust corrections are important, this number is a lower limit, and 
the actual star formation rates in dust lane early-type galaxies may be even higher. 

Figure [9]& shows the distribution of the black hole accretion rate, in Eddington units. 
Bolometric luminosity is derived using the O [III] line luminosity, and the usual correction 



Dust lane galaxies as tracers of gas-rich mergers 17 



All ETGs 
Dust lane ETGs 




-1 1 

log SFRsoss / Ms yr"' 

(a) star formation rate 



0.25 : 

0.20 ; 
0.15 ■ 

0.10 : 

0.05 ; 

0.00 : 



All ETGs ■ 
Dust lone ETGs 




-4-2 2 4 

log L„ /Udd 

(b) rhBH 



Figure 9. Top: Distribution of star formation rates for dust lane early-type galaxies and a matched control sample of all 
early-types. Both samples have 100 < t2 < 400 Myrs. Bottom: Distribution of ttibh in Eddington units. Median values are 
again shown by dashed lines. Dust lane early-type galaxies have significantly higher star formation and black hole accretion 
rates than the control sample, consistent with these objects being associated with recent gas-rich mergers. 



factor of 3500 (IHeckman et al. 2004|1 is applied. The black hole mass is estimated via its 
velocity dispersion fITremaine et al. 2002^ . Dust lane early- types show accretion rates that 
are again a factor of three higher than the control sample, consistent with the discrepancy 
in the star formation rates. 

A final confirmation of this picture comes from emission-line AGN fractions in the two 
samples. Almost all (84 percent) dust lane early-types are classified as either Seyfert, LINER 
or transition objects, compared to only 24 percent for the control sample. More than 60 
percent of the objects in the control sample have emission lines that are too weak to place 
them on the BPT diagnostic, suggesting that these objects undergo very low-level (if any) 
star formation and AGN activity, despite having relatively young starbursts. This is in stark 
contrast to dust lane hosts with the same starburst ages, and confirms that the dust features 
are simply associated with galaxies that have acquired large amounts of gas, most likely in 
a recent merger. 



5 CONCLUSIONS 

We consider the origin and evolution of dust lane early-type galaxies. In a companion paper, 
we showed that these objects are preferentially located in poor environments, are morpholog- 
ically disturbed, and have higher star formation rates than the mean for early-type galaxies. 
In this work we addressed the issue of the origin of these objects. By employing optical 
(SDSS) and radio (FIRST and NVSS) surveys we examined AGN properties of dust lane 
early-type galaxies and a relevant early-type control sample. Our findings are as follows. 



18 Shabala et al. 

1. Dust lane early- types are much more likely to host emission-line AGN than the control 
sample of early-type galaxies. 

2. Radio-loud AGN in dust lane early-types do not show an environmental dependence, 
with similar fractions detected in the field and groups/clusters. By contrast, control sample 
early-type galaxies are much more likely to be radio-loud if located in a cluster. 

3. Radio- loud AGN in dust lane early-types are much more likely to also be detected in 
emission lines than radio-loud AGN in the control sample. 

These results all strongly suggest that minor mergers act as a trigger for AGN activity 
in dust lane early-type galaxies. 
Furthermore, 

4. Dust lane early-types have starburst ages that are younger than the mean for early- 
type galaxies. 

5. Dust lane early-type galaxies exhibit higher star formation and black hole accretion 
rates than the control sample early-types with the same starburst ages. 

These results indicate that the minor mergers are both gas-rich and recent. The mergers 
act as a trigger for both the starburst and AGN activity. The dust lane phase is associated 
with starburst galaxies in which the AGN has already been triggered, but has not yet shut 
off star formation completely. 



ACKNOWLEDGEMENTS 

S.S. thanks the Australian Research Council and New College, Oxford for research fellow- 
ships. Y.S.T. is grateful to the French Ministry of Foreign and European Affairs for an 
Egide- Eiffel scholarship, and Ecole Polytechnique. S.K. acknowledges a Research Fellowship 
from the Royal Commission for the Exhibition of 1851, an Imperial College Junior Research 
Fellowship and a Senior Research Fellowship from Worcester College, Oxford. 

We are grateful to Martin Bureau, Christophe Pichon and Adrianne Slyz for illuminating 
discussions, the referee for useful comments, and Sally Hales for assistance with FIRST and 
NVSS data. 

This work would not have been possible without the contributions of citizen scientists 
as part of the Galaxy Zoo 2 project. 

Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan Foundation, 
the Participating Institutions, the National Science Foundation, the US Department of En- 



Dust lane galaxies as tracers of gas-rich mergers 19 

ergy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, 
the Max Planck Society and the Higher Education Funding Council for England. The SDSS 



web site is http://www.sdss.org 



The SDSS is managed by the Astrophysical Research Consortium for the Participating 
Institutions. The Participating Institutions are the American Museum of Natural History, 
Astrophysical Institute Potsdam, University of Basel, University of Cambridge, Case West- 
ern Reserve University, University of Chicago, Drexel University, Fermilab, the Institute 
for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint 
Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmol- 
ogy, the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos 
National Laboratory, the Max-Planck- Institute for Astronomy (MPIA), the Max-Planck- 
Institute for Astrophysics (MPA), New Mexico State University, Ohio State University, 
University of Pittsburgh, University of Portsmouth, Princeton University, the United States 
Naval Observatory and the University of Washington. 

The NVSS and FIRST surveys were carried out using the National Radio Astronomy 
Observatory VLA. The National Radio Astronomy Observatory is a facility of the National 
Science Foundation operated under cooperative agreement by Associated Universities, Inc. 

REFERENCES 

Allen, S. W., Dunn, R. J. H., Fabian A. C, Taylor, G. B., Reynolds, C. S. 2006, MNRAS 372, 21 

Baldwin, J. A., Phillips, M. M., Terlevich, R. 1981, PASP 93, 817 

Beeker, R. H., White, R. L., Helfand, D. J. 1995, ApJ 450, 559 

Bell, E. F., Melntosh, D. H., Katz, N., Weinberg, M. D. 2003, ApJS 149, 289 

Best, P. N., Kauffmann, G., Heckman, T. M., Ivezic, Z. 2005a, MNRAS 362, 9 

Best, P. N., Kauffmann, G., Heckman, T. M., Brinchmann, J., Chariot, S., Ivezic, Z., White, S. D. M. 2005b, MNRAS 362, 25 

Binney J. J., Tabor, G. R. 1995, MNRAS 276, 663 

Brinchmann, J., Chariot, S., White, S. D. M., Tremonti, C, Kauffmann, G., Heckman, T., Brinkmann, J. 2004, MNRAS 351, 

1151 
Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F., Perley R. A., Taylor, G. B., Broderick J. J. 1998, AJ 115, 1693 
Davies, R. I., Sanchez, F. M., Genzel, R., Tacconi, L. J., Hicks, E. K. S., Friedrich, S., Sternberg, A. 2007, ApJ 671, 1388 
Draine, B. T. 2009, Astronomical Society of the Pacific Conf. Series 414, 453 
Hardcastle, M. J., Evans, D. A., Croston, J. H. 2007, MNRAS 376, 1849 
Haring, N., Rix, H.-W. 2004, ApJ 604, 89 

Heckman, T. M., Kauffmann, G., Brinchmann, J., Chariot, S., Tremonti, C, White, S. D. M. 2004, ApJ 613, 109 
Kaiser, C. R., Alexander P. 1997, MNRAS 286, 215 

Kaiser, C. R., Dennett-Thorpe, J., Alexander, P. 1997, MNRAS 286, 215 
Kauffmann, G., Heckman, T. M., Tremonti, C, et al. 2003c, MNRAS 346, 1055 
Kaviraj, S. 2010, MNRAS 406, 382 



20 Shabala et al. 

Kaviraj, S. 2010, MNRAS 408, 170 

Kaviraj, S., Ting, Y. S., Shabala, S. S, et al. 2012, MNRAS submitted (Paper 1) 

Kennicutt, R. C. Jr. 1998, ApJ 498, 541 

Krause, M. G. H., Gaibler, V. 2009. larXiv:0906.2122l 

Magorrian, J., Trcmaine, S., Richstone, D., et al. 1998, AJ 115, 2285 

Martini, P., Regan, M. W., Mulchaey, J. S., Pogge, R. W. 2003, ApJ 589, 774 

Meier, D. L. 2001, ApJ Lett. 548, 9 

Rawlings, S., Saunders, R. 1991, Nature 349, 138 

Schawinski, K., Thomas, D., Sarzi, M. et al. 2007, MNRAS 382, 1415 

Schawinski, K., Lintott, C. J., Thomas, D. et al. 2009, MNRAS 396, 818 

Schawinski, K., Lintott, C. J., Thomas, D. et al. 2009, ApJ 690, 1672 

Shabala, S. S., Ash, S. A., Alexander, P., Riley, J. M. 2008, MNRAS 388, 625 

Shabala, S. S., Alexander, P. 2009, ApJ 699, 525 

Shakura, N. L, Sunyaev, R. A. 1973, A&A 24, 337 

Simoes Lopes, R. D., Storchi-Bergmann, T., de Fatima Saraiva, M., Martini, P. 2007, ApJ 655, 718 

Tremaine, S., Gebhardt, K., Bender, R. et al. 2002, ApJ 574, 740 

Tremonti, C. A., Heckman, T. M., Kauffmann, G. et al. 2004, ApJ 613, 898 

Yang, X., Mo, H. J., van den Bosch, F. C., Pasquah, A., Li, C., Barden, M. 2007, ApJ 671, 153 

Yun, M. S., Reddy, N. A., Condon, J. J. 2001, ApJ 554, 803