LETTER doi:10.

1038/nature09694

Supermassive black holes do not correlate with

galaxy disks or pseudobulges
John Kormendy1, R. Bender2,3 & M. E. Cornell1

The masses of supermassive black holes are known to correlate with found observationally but is now understood to be a result of fun-
the properties of the bulge components of their host galaxies1–5. In damentally different formation mechanisms6.
contrast, they seem not to correlate with galaxy disks1. Disk-grown Classical bulges (red points in Figs 1 and 2) are indistinguishable
‘pseudobulges’ are intermediate in properties between bulges and from ellipticals in their structure, velocity distributions and parameters.
disks6; it has been unclear whether they do1,5 or do not7–9 correlate Our well-developed picture is that they formed by galaxy mergers13 in
with black holes in the same way that bulges do. At stake in this our hierarchically clustering universe. Mergers are discrete events that
issue are conclusions about which parts of galaxies coevolve with are separated by long ‘dead times’. They occur on short timescales,
black holes10, possibly by being regulated by energy feedback from approximately equal to the crossing time of the merging galaxies.
black holes11. Here we report pseudobulge classifications for Gravitational torques scramble disks into ellipticals13 and dump large
galaxies with dynamically detected black holes and combine them quantities of gas into the centre. Observations14 and theory15 suggest
with recent measurements of velocity dispersions in the biggest that the process feeds both starbursts and black holes and causes the
bulgeless galaxies12. These data confirm that black holes do not latter to grow rapidly in quasar-like events.
correlate with disks and show that they correlate little or not at Pseudobulges6 (blue points in Figs 1 and 2) are observed to be more
all with pseudobulges. We suggest that there are two different disk-like than classical bulges. They are believed to form more gently by
modes of black-hole feeding. Black holes in bulges grow rapidly the gradual internal redistribution of angular momentum in quiescent
to high masses when mergers drive gas infall that feeds quasar-like galaxy disks. The driving agents are non-axisymmetries such as bars.
events. In contrast, small black holes in bulgeless galaxies and in One result is the gradual build-up of a high-density central component
galaxies with pseudobulges grow as low-level Seyfert galaxies. that can be recognized (Supplementary Information) because it remains
Growth of the former is driven by global processes, so the biggest disk-like. These components are called ‘pseudobulges’ to emphasize
black holes coevolve with bulges, but growth of the latter is driven their different formation mechanism without forgetting that they look
locally and stochastically, and they do not coevolve with disks and superficially like—and are commonly confused with—classical bulges.
pseudobulges. The difference from bulges that is most relevant here is that the gradual
In Fig. 1b, we show the well-known correlation1,5 between dynam- gas infall that builds pseudobulges may concurrently provide less black-
ically measured black-hole masses, M$, and the absolute magnitudes hole feeding and thus may drive slower black-hole growth. One purpose
of elliptical galaxies (Fig. 1b, black points) and the bulge parts of disk of this paper is to contrast black hole/bulge and black hole/pseudobulge
galaxies (Fig. 1b, red points). The correlation has x2 5 12 per degree of correlations to look for clues about black-hole growth mechanisms and
freedom, implying moderate intrinsic scatter. This result and a tighter the consequent coevolution (or not) of black holes with host galaxies.
correlation2–5 between M$ and host velocity dispersion, s (Fig. 2), With this context, we can interpret Fig. 1. Galaxies that contain
motivate the idea that black holes and host bulges evolve together classical bulges are consistent with the correlations for elliptical galaxies
and regulate each other10,11. All new results in this paper stand in except for one discrepant object (the bulge-dominated S0 galaxy NGC
contrast to these two correlations. 4342). The implication is that classical bulges and ellipticals coevolve
In Fig. 1a, we plot M$ versus the K-band absolute magnitude, MK, with black holes in the same way. For that coevolution, it is irrelevant
of only the disk part of the host galaxy, with the bulge luminosity that bulges are now surrounded by disks whereas ellipticals are not.
removed. We conclude that black holes do not correlate with galaxy We reach a different conclusion for pseudobulges on the basis of new
disks. This confirms an earlier conclusion1 based on the more indirect classifications and measurements of pseudobulge-to-total luminosity
observation that black holes do not correlate with total (bulge plus ratios for all disk galaxies in our sample5 that have dynamical black-hole
disk) luminosities of disk galaxies. In Fig. 1a, the colour of each point— detections (J.K., manuscript in preparation; see Supplementary
which indicates bulge type (see below)—is irrelevant. A least-squares Information for a list). A conservative interpretation of Fig. 1b is that
fit to the red and blue points has correlation coefficient r 5 0.41. Errors pseudobulges are roughly consistent with the correlation for classical
at 1 s.d. imply x2 5 81 per degree of freedom: the data do not respect bulges and ellipticals but have much more scatter. In particular, some
the weak anticorrelation. The green points for pure-disk (that is, com- deviate from the correlation for ellipticals in having smaller black holes.
pletely bulgeless) galaxies further confirm the large scatter and lack of This was not seen in some previous work1,5 because samples were small
correlation. and because many pseudobulges in black-hole galaxies had not been
In Fig. 1c, we plot M$ versus MK for pseudobulge components with classified. But, as published samples have grown larger, the hints have
disk light removed. They are also included in Fig. 1b (light blue) to grown stronger that pseudobulges do not correlate with black holes in
show how they compare with classical bulges and elliptical galaxies. the same way as classical bulges7–9. We confirm these hints. Particularly
Pseudobulges required explanation, as follows. compelling is the fact that our galaxies and a new sample of black-hole
Much work over several decades has shown that the high-stellar- detections based on water masers9 have no overlap and independently
density central components in disk galaxies—all of which used to be lead to the same conclusion.
called ‘bulges’—come in two varieties. How to distinguish them is Figure 1c shows the pseudobulges without guidance from the red
discussed in Supplementary Information. The difference was first and black points. The sample is small, but we have enough dynamic
1
Department of Astronomy, University of Texas at Austin, 1 University Station, Austin, Texas 78712-0259, USA. 2Max-Planck-Institut für Extraterrestrische Physik, Giessenbachstrasse, D-85748 Garching-
bei-München, Germany. 3Universitäts-Sternwarte, Scheinerstrasse 1, D-81679 München, Germany.

3 7 4 | N AT U R E | VO L 4 6 9 | 2 0 J A N U A RY 2 0 1 1
©2011 Macmillan Publishers Limited. All rights reserved
 LETTER RESEARCH

a a 10
10
Disk MK Bulges and ellipticals
9

log(M /M )
8
log(M /M )

6 7

6
4

b b 10
10
Bulge MK Pseudobulges
9
9
NGC 4342

log(M /M )
8
log(M /M )

7
7

6
6

c 5
10 1.5 2.0 2.5
Pseudobulge MK log[σ (km s–1)]

9 Figure 2 | Correlation of dynamically measured black-hole masses with the

velocity dispersions of their host galaxies. Black points are for elliptical
galaxies, red points are for classical bulges, blue points are for pseudobulges and
log(M /M )

8 the green point is for a nuclear star cluster. Data sources are given in
Supplementary Information. Error bars, 1 s.d. The red and black points show
the well known M$–s correlation2–5: x2 5 5.0 per degree of freedom and
7 r 5 0.89. Reducing x2 to 1.0 implies that the intrinsic scatter in log(M$/M[) at
fixed s is 0.26, consistent with previous derivations4,5. This is the tightest
correlation between black holes and host galaxy properties and the one that
6 most motivates the idea that black holes and bulges coevolve. In contrast, the
blue points for pseudobulges show no correlation: x2 5 10.4 and r 5 20.08.
This extends suggestions7–9 that the M$–s relation for pseudobulges is
–18 –20 –22 –24 –26 different from that for classical bulges and elliptical galaxies.
MK
range to conclude that we see no correlation at all. The cumulative
Figure 1 | Correlations of dynamically measured black-hole masses with the amount of black-hole growth is not extremely different in classical
luminosities of different parts of their host galaxies. Here MK is the K-band bulges and pseudobulges, but there is no sign in the correlations that
(2.2-mm) absolute magnitude of the disk component with bulge light removed black-hole feeding has affected the pseudobulges.
(a), of the bulge with disk light removed (b) and of the pseudobulge with disk The second and more compelling black hole/host galaxy correlation
light removed (c). All plotted data are published elsewhere; parameters and is the one between M$ and the velocity dispersion, s, of the stars at
sources are discussed in Supplementary Information, and those for disk galaxies radii where they do not feel the black hole gravitationally2–5. Here s is
are tabulated there. Elliptical galaxies are plotted in black, classical bulges are averaged inside the ‘effective radius’, re, that contains half of the bulge
plotted in red and pseudobulges are plotted in dark blue. One galaxy with a
light. Figure 2 shows this correlation.
dominant pseudobulge but with a possible small classical bulge (NGC 2787) is
plotted with a blue symbol that has a red centre. In least-squares fits, it is
As is well known, ellipticals and classical bulges share the same tight
included with the pseudobulges. Error bars, 1 s.d. In b, the red and black points correlation. But as in Fig. 1, pseudobulges at best show a much larger
show a good correlation between M$ and bulge luminosity: a symmetric, least- scatter (Fig. 2a). Without the guidance of the red and black points
squares fit4 of a straight line has x2 5 12.1 per degree of freedom and a Pearson (Fig. 2b), they show essentially no correlation. Larger samples that
correlation coefficient of r 5 20.82. (All x2 values quoted in this paper are per reach smaller values of M$ may show a weak relationship16–20. But
degree of freedom.) In contrast, in a the red and blue points together confirm a we conclude that classical bulges and pseudobulges show very different
previous result1 that black holes do not correlate with disks: x2 5 81 and correlations with M$. Those for classical bulges are tight enough to
r 5 0.41. Green points are for galaxies that contain neither a classical bulge nor a suggest coevolution. Whether pseudobulges correlate with M$ with
pseudobulge but only a nuclear star cluster, that is, pure-disk galaxies. They are large scatter or not at all, the weakness of any correlation (r 5 20.08
not included in the above fit, but they strengthen our conclusion. Similarly, in
here) makes no compelling case that pseudobulges and black holes
c the blue points for pseudobulges show no correlation: x2 5 63 and r 5 0.27. In
all panels, galaxies that have only M$ limits are plotted with open symbols; they coevolve, beyond the obvious expectation that it is easier for bigger
were chosen to increase our dynamic range. They too support our conclusions. black holes and bigger pseudobulges to grow in bigger galaxies that
This figure uses K-band magnitudes to minimize effects of star formation and contain more fuel.
internal absorption, but in Supplementary Information we show that Fig. 1 From the point of view of galaxy formation by hierarchical cluster-
looks essentially the same for V-band (0.55-mm) magnitudes. ing, pseudobulge galaxies are already pure-disk galaxies12. Even more
2 0 J A N U A RY 2 0 1 1 | VO L 4 6 9 | N AT U R E | 3 7 5
©2011 Macmillan Publishers Limited. All rights reserved
 RESEARCH LETTER

extreme galaxies that contain neither a classical bulge nor a pseudo- 8. Nowak, N. et al. Do black hole masses scale with classical bulge luminosities only?
The case of the two composite pseudo-bulge galaxies NGC 3368 and NGC 3489.
bulge can, in some cases, also contain black holes. Some have active Mon. Not. R. Astron. Soc. 403, 646–672 (2010).
galactic nuclei (AGNs), although these are rare in bulgeless galaxies21. 9. Greene, J. E. et al. Precise black hole masses from megamaser disks: black hole–
And black holes with M$ 5 104M[–106M[ (M[, solar mass) have bulge relations at low mass. Astrophys. J. 721, 26–45 (2010).
10. Ho, L. C. (ed.) Coevolution of Black Holes and Galaxies (Carnegie Observatories
been discovered with confidence in bulgeless (and pseudobulgeless) Astrophys. Ser. 1, Cambridge Univ. Press, 2004).
galaxies18,21. The most extreme example is NGC 4395, an Sm galaxy 11. Silk, J. & Rees, M. J. Quasars and galaxy formation. Astron. Astrophys. 331, L1–L4
that contains only a tiny, globular-cluster-like nucleus but that has an (1998).
12. Kormendy, J., Drory, N., Bender, R. & Cornell, M. E. Bulgeless giant galaxies
AGN powered by a black hole22 with M$ 5 (3.6 61.1) 3 105M[ well challenge our picture of galaxy formation by hierarchical clustering. Astrophys. J.
measured by reverberation mapping23. It is the green point with the 723, 54–80 (2010).
small error bar in Fig. 2. Other such objects include the Sd galaxy NGC 13. Toomre, A. in Evolution of Galaxies and Stellar Populations (eds Tinsley, B. M. &
362124 and the spheroidal galaxy POX 52, which contains an AGN Larson, R. B.) 401–426 (Yale Univ. Observatory, 1977).
14. Sanders, D. B. et al. Ultraluminous infrared galaxies and the origin of quasars.
powered by a black hole of mass M$ < 105M[ (refs 18, 25). Also, it is Astrophys. J. 325, 74–91 (1988).
likely that low-M$ AGN samples include bulgeless galaxies16–19, 15. Hopkins, P. F. et al. A unified, merger-driven model of the origin of starbursts,
although some are distant and not well resolved. quasars, the cosmic X-ray background, supermassive black holes, and galaxy
spheroids. Astrophys. J. Suppl. Ser. 163, 1–49 (2006).
The lack of correlation of black holes with disks and pseudobulges 16. Barth, A. J., Greene, J. E. & Ho, L. C. Dwarf Seyfert 1 nuclei and the low-mass end of
plus the discovery of black holes in bulgeless galaxies together are the MBH–s relation. Astrophys. J. 619, L151–L154 (2005).
critical clues to black-hole feeding mechanisms. They motivate the 17. Greene, J. E. & Ho, L. C. The MBH–s* relation in local active galaxies. Astrophys. J.
641, L21–L24 (2006).
following hypothesis: we suggest that there are two fundamentally 18. Greene, J. E., Ho, L. C. & Barth, A. J. Black holes in pseudobulges and spheroidals: a
different feeding mechanisms for black holes. change in the black hole–bulge scaling relations at low mass. Astrophys. J. 688,
The traditional mechanism is rapid feeding during major mergers, 159–179 (2008).
19. Bentz, M. C., Peterson, B. M., Pogge, R. W. & Vestergaard, M. The black hole mass–
when large amounts of gas fall into galaxy centres. We follow sugges- bulge luminosity relationship for active galactic nuclei from reverberation mapping
tions14,15 that, at some, perhaps late, stage in the merger, the black hole and Hubble Space Telescope imaging. Astrophys. J. 694, L166–L170 (2009).
grows quickly in a quasar-like event. These are the growth episodes 20. Woo, J.-H. et al. The Lick AGN monitoring project: the MBH–s* relation for
reverberation-mapped active galaxies. Astrophys. J. 716, 269–280 (2010).
that dominate the waste mass budget of quasar activity15,26 that is well 21. Ho, L. C. Nuclear activity in nearby galaxies. Annu. Rev. Astron. Astrophys. 46,
explained by the observed density of the biggest black holes in the 475–539 (2008).
universe27. In this mode, black-hole and galaxy growth are controlled 22. Filippenko, A. V. & Ho, L. C. A low-mass central black hole in the bulgeless Seyfert 1
by the same global processes, and these, we suggest, result in black galaxy NGC 4395. Astrophys. J. 588, L13–L16 (2003).
23. Peterson, B. M. et al. Multiwavelength monitoring of the dwarf Seyfert 1 galaxy NGC
hole/bulge coevolution. If black holes and galaxy formation ever regu- 4395. I. A reverberation-based measurement of the black hole mass. Astrophys. J.
late each other11, this is the likely setting in which it happens. 623, 799–808 (2005).
In contrast, the nuclear activity in bulgeless galaxies is, by and large, 24. Barth, A. J., Strigari, L. E., Bentz, M. C., Greene, J. E. & Ho, L. C. Dynamical constraints
on the masses of the nuclear star cluster and black hole in the late-type spiral
weak21. Big pseudobulges (NGC 1068 is the highest-luminosity one in galaxy NGC 3621. Astrophys. J. 690, 1031–1044 (2009).
Fig. 1) can host classical Seyfert nuclei, but there is no sign that these 25. Thornton, C. E., Barth, A. J., Ho, L. C., Rutledge, R. E. & Greene, J. E. The host galaxy
affect global galaxy structure. NGC 1068 is a prototypical oval galaxy6 and central engine of the dwarf active galactic nucleus POX 52. Astrophys. J. 686,
892–910 (2008).
with a prominent pseudobulge that, we believe, formed slowly by 26. Hopkins, P. F. & Hernquist, L. Fueling low-level AGN activity through stochastic
inward gas transport in spite of episodic nuclear activity. We suggest accretion of cold gas. Astrophys. J. Suppl. Ser. 166, 1–36 (2006).
that the second mode of black-hole growth is such weak nuclear activity 27. Yu, Q. & Tremaine, S. Observational constraints on growth of massive black holes.
Mon. Not. R. Astron. Soc. 335, 965–976 (2002).
driven stochastically by local processes that feed gas from ,102 pc, 28. Schawinski, K. et al. Galaxy zoo: the fundamentally different co-evolution of
where it make a pseudobulge, in to the black hole. The processes are supermassive black holes and their early- and late-type host galaxies. Astrophys. J.
not understood in detail. But exactly this kind of feeding mode is pro- 711, 284–302 (2010).
posed and modelled in ref. 26; the models suggest that this kind of 29. Kumar, P. & Johnson, J. L. Supernovae-induced accretion and star formation in the
inner kiloparsec of a gaseous disc. Mon. Not. R. Astron. Soc. 404, 2170–2176 (2010).
feeding does not affect galaxy formation. Differences in black-hole 30. Hopkins, P. F. & Quataert, E. How do massive black holes get their gas? Mon. Not. R.
growth and coevolution with host galaxies have also been proposed in Astron. Soc. 407, 1529–1564 (2010).
studies of the lowest-M$ black holes detected mainly through AGN Supplementary Information is linked to the online version of the paper at
activity16,18. The same picture can also be reached by studying AGN www.nature.com/nature.
demographics28. Other recent papers29,30 further explore local processes Acknowledgements We acknowledge with pleasure our collaboration with N. Drory on
of black-hole feeding. Figures 1 and 2 tell us that this mode involves little work leading up to this paper. We thank N. Drory and J. Greene for helpful comments on
or no coevolution of black holes with any component of the host galaxy. the manuscript and J. Greene for communicating the maser black-hole detection
results before publication. We also thank K. Gebhardt for permission to use M$ values
The seed black holes that grew into today’s supermassive black holes for NGC 4736 and NGC 4826, and J. Jardel for permission to use his updated M$ value
are not securely identified. But the smallest black holes that are grown by for NGC 4594 before publication. Some data used here were obtained with the Hobby–
means of local feeding plausibly remain most like those seeds. Because Eberly Telescope (HET), which is a joint project of the University of Texas at Austin,
Pennsylvania State University, Stanford University, Ludwig-Maximilians-Universität
mergers of disk galaxies are believed to make bulges and ellipticals, we Munich and Georg-August-Universität Göttingen. It is named in honour of its principal
suggest that the small black holes that are grown by local processes are the benefactors, W. P. Hobby and R. E. Eberly. We made extensive use of data from the Two
seeds of the generally larger black holes that are grown in part by mergers. Micron All Sky Survey, a joint project of the University of Massachusetts and the Infrared
Processing and Analysis Center (IPAC)/California Institute of Technology funded by
Received 12 July; accepted 19 November 2010. NASA and by the National Science Foundation (NSF). We also made extensive use of
the NASA/IPAC Extragalactic Database (NED), which is operated by California Institute
1. Kormendy, J. & Gebhardt, K. in Proc. 20th Texas Symp. Relativ. Astrophys. (eds of Technology and the Jet Propulsion Laboratory under contract with NASA; of the
Wheeler, J. C. & Martel, H.) 363–381 (American Institute of Physics, 2001). HyperLeda database (http://leda.univ-lyon1.fr); and of the NASA Astrophysics Data
2. Ferrarese, L. & Merritt, D. A fundamental relation between supermassive black System bibliographic services. Finally, we are grateful to the NSF for grant support.
holes and their host galaxies. Astrophys. J. 539, L9–L12 (2000). Author Contributions J.K. led the programme, carried out the analysis for this paper
3. Gebhardt, K. et al. A relationship between nuclear black hole mass and galaxy and wrote most of the text. M.E.C. oversaw the HET observations, preprocessed the HET
velocity dispersion. Astrophys. J. 539, L13–L16 (2000). spectra and provided technical support throughout the project. R.B. calculated the
4. Tremaine, S. et al. The slope of the black hole mass versus velocity dispersion velocity dispersions from the HET spectra and made all least-squares fits. All authors
correlation. Astrophys. J. 574, 740–753 (2002). contributed to the writing of the paper.
5. Gültekin, K. et al. The M–s and M–L relations in galactic bulges, and determinations
of their intrinsic scatter. Astrophys. J. 698, 198–221 (2009). Author Information Reprints and permissions information is available at
6. Kormendy, J. & Kennicutt, R. C. Secular evolution and the formation of www.nature.com/reprints. The authors declare no competing financial interests.
pseudobulges in disk galaxies. Annu. Rev. Astron. Astrophys. 42, 603–683 (2004). Readers are welcome to comment on the online version of this article at
7. Hu, J. The black hole mass–stellar velocity dispersion correlation: bulges versus www.nature.com/nature. Correspondence and requests for materials should be
pseudo-bulges. Mon. Not. R. Astron. Soc. 386, 2242–2252 (2008). addressed to J.K. ([email protected]).

3 7 6 | N AT U R E | VO L 4 6 9 | 2 0 J A N U A RY 2 0 1 1
©2011 Macmillan Publishers Limited. All rights reserved