The Fueling and Evolution of AGN: Internal and External Triggers S. Jogee Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA [email protected] 1 Introduction Thequestforacoherentpictureofnuclearactivityhaswitnessedgiantleaps in the last decades. Four decades ago, the idea was put forward that accre- tion of matter onto a massive compact object or a supermassive black hole (SMBH) of mass >106 M(cid:1) could power very luminous active galactic nu- clei (AGN), in particular, quasi-stellar objects (QSOs) (Lynden-Bell 1969; Soltan 1982; Rees 1984). In the last decade, dynamical evidence increasingly suggests that SMBH pervade the centers of most massive galaxies (Sect. 2 and references therein). The challenge has now shifted towards probing the fueling and evolution of AGN over a wide range of cosmic lookback times, and elucidating how they relate to their host galaxies in both the local and cosmological context. Inthisreview,IwillfocusonthefuelingandevolutionofAGNunderthe influence of internal and external triggers. In the nature versus nurture par- adigm, I use the term internal triggers to refer to intrinsic properties of host galaxies (e.g., morphological or Hubble type, color, and non-axisymmetric features such as large-scale bars and nuclear bars) while external triggers refer to factors such as environment and interactions. The distinction is an over-simplificationasmanyofthesocalledintrinsicpropertiesofgalaxiescan be induced or dissolved under the influence of external triggers. Connections will be explored between the nuclear and larger-scale properties of AGN, both locally and at intermediate redshifts. One of the driving objectives is to understand why not all relatively massive galaxies show signs of AGN ac- tivity (via high-excitation optical lines or X-ray emission) despite mounting dynamical evidence that they harbor SMBHs. The most daunting challenge in fueling AGN is arguably the angular momentum problem (Subsect. 3.2). Even matter located at a radius of a few hundred pc must lose more than 99.99% of its specific angular momentum before it is fit for consumption by a BH. The sequence of this review is as follows. Section 2 briefly addresses BH demographics and the BH-bulge-halo correlations. Section 3 sets the stage for the rest of this paper by providing an overview of central issues in the S. Jogee: The Fueling and Evolution of AGN: Internal and External Triggers, Lect. Notes Phys.693,143–183(2006) DOI10.1007/3-540-34621-X6 (cid:2)c Springer-VerlagBerlinHeidelberg2006 144 S. Jogee fueling of AGN and circumnuclear starbursts. In particular, I review mass accretion rates, angular momentum requirements, the effectiveness of differ- entfuelingmechanisms,andthegrowthandmassdensityofBHsatdifferent epochs.ThesecentralissuesinSect.3areattackedinmoredetailinSects.4–9 which describe different fueling mechanisms including mergers and interac- tions(Sect.5),large-scalebars(Sect.6),nuclearbars(Sect.7),nuclearspirals (Sect. 8), and processes relevant on hundred pc to sub-pc scales (Sect. 9). I conclude with a summary and future perspectives in Sect. 10. Complemen- tary reviews on mass transfer and central activity in galaxies include those by Shlosman (2003), Combes (2003), Knapen (2004), and Martini (2004). 2 BH Demographics and BH-Bulge-Halo Correlations 2.1 Measurement of BH Masses ThetermSMBHsreferstoBHshavingmassesMbh >106 M(cid:1),incontrastto intermediate mass BHs (IMBHs) with Mbh ∼ 102–106 M(cid:1), and stellar mass BHs.PropertiesofSMBHsaregenerallystudiedthroughaccretionsignatures ofBHsortheirgravitationalinfluence.Thestrongestdynamicalevidencefor SMBHs are in our Galaxy and in NGC 4258. In these systems, the large central densities inferred within a small resolved radius can be accounted for by a SMBH, but not by other possibilities such as collections of compact objects, star clusters, or exotic particles. In our Galaxy, proper motion mea- surements set stringent constraints on the central potential (Sch¨odel et al. 2003; Ghez et al. 2003; Genzel et al. 2000), yielding Mbh ∼ 3–4 × 106 M(cid:1). In NGC 5248, VLBA maser observations reveal Keplerian motions implying Mbh ∼ 3.9×107 M(cid:1) (Miyoshi et al. 95). In the last decade, high resolution gas and stellar dynamical measure- ments from ground-based (e.g., Kormendy & Richstone 1995) and HST ob- servations(e.g.,Harmsetal.1994;Ferrareseetal.1996;vanderMarel&van den Bosch 1998; Ferrarese & Ford 99; Gebhardt et al. 2000) have provided compelling evidence that several tens of galaxies host massive central dark objects (CDOs) which are likely to be SMBHs. The more reliable dynamical measurementstendtobefromobservationswhichresolvetheradiusofinflu- ence (Rg−bh) within which the gravitational force of the BH exceeds that of nearby stars with velocity dispersion σ, namely, (cid:10) (cid:11)(cid:8) (cid:9) GM M σ −2 Rg−bh = σ2bh =11.2 pc 108 bMh(cid:1) 200 km s−1 (1) However, the scales probed by these measurements are still several 105– 106 times the Schwarzschild radius (Rs−bh) of the BH, namely, (cid:10) (cid:11) 2GM M Rs−bh = c2bh =5×10−4 pc 108 bMh(cid:1) (2) Fueling AGN and Starbursts 145 The majority of the afore mentioned reliable measurements target ellip- ticals and a few early-type (Sa-Sbc) spirals with central σ < 60 km s−1, and probeBHmassesintherange107–109M(cid:1).Conversely,measuringBHmasses in late-type spirals and dwarf galaxies poses many challenges, and there are no firm measurements of BH masses below 106 M(cid:1). However, theoretical models and a mounting body of observational evidence put the existence of IMBHs on a relatively firm footing (see review by van der Marel 2003). The first challenge in measuring the masses of IMBHs is that the gravita- tional radii of such BHs are typically too small to be easily resolved even with HST. A second complication is that late-type spirals and dwarf galax- ies which might harbor such BHs also tend to host a bright 106–107 M(cid:1) stellar cluster (Boker et al. 1999) whose dynamical effect can mask that of the BH. A 104–105 M(cid:1) BH (Filipenko & Ho 2003) has been invoked in the Sm dwarf NGC 4395 which hosts the nearest and lowest luminosity Seyfert 1 nucleus. Upper limits on BH masses are reported in several systems, e.g., 106–107 M(cid:1) for six dwarf ellipticals in Virgo (Geha, Guhathakurta, & van der Marel 2002), 5×105 M(cid:1) for the Scd spiral IC342 (Boker et al.1999). Gebhardt,Rich,&Ho(2002)inferthepresenceofanIMBHwithamassofa few×104 M(cid:1) inoneofthemostmassivestellarclusters(G1)inM31,butan alternative interpretation of the dataset has been presented by Baumgardt et al. (2003). A tantalizing dark central mass concentration of a few ×103 M(cid:1) (Gerssen et al. 2003) is reported in the globular cluster M15 from HST data, but it remains unclear whether it is an IMBH. Chandra observations of ultraluminous X-ray sources also suggest the presence of IMBHs (Clobert & Miller 2004 and references therein). At many levels, measuring BH masses in local AGN such as Seyferts and LINERS is more challenging than corresponding measurements in massive quiescent galaxies. The bright non-thermal active nucleus in Seyfert galaxies can drown the spectroscopic features from which dynamical measurements aremade.Consequently,BHmassesinlocalAGNarecommonlymappedwith alternative techniques such as reverberation mapping (Blandford & McKee 1982;Peterson1993;seePetersontheseproceedings)whereoneestimatesthe virial mass inside the broad-line region (BLR) by combining the velocity of the BLR with an estimate of the size of the BLR based on time delay mea- surements. Reverberation mapping can typically probe scales ∼600 Rs−bh and has yielded BH masses for several tens of AGN (Peterson 1993; Wan- del, Peterson, & Malkan 1999; Kaspi 2000). Earlier controversies existed on the reliability of the method due to purported systematic differences in the BH-to-bulge mass ratio between AGN with reverberation mapping data and quiescentgalaxiesorQSOs.However,recentwork(e.g.,Ferrareseetal.2001) claims that for AGN with accurate measurements of stellar velocity disper- sions,thereverberationmassesagreewiththeBHmassdeterminedfromthe tight M –σ relation (Subsect. 2.2) which is derived from quiescent galaxies. bh 146 S. Jogee Fig.1. CorrelationbetweencentralBHmassandcircumnuclearvelocity dispersion – Black hole mass versus bulge luminosity (left) and the luminosity- weighted aperture dispersion within the effective radius (right). Green squares de- notegalaxieswithmaserdetections,redtrianglesarefromgaskinematics,andblue circlesarefromstellarkinematics.Solidanddottedlinesarethebest-fitcorrelations and their 68% confidence bands. (From Gebhardt et al. 2000) 2.2 Relationship of the Central BH to the Bulge and Dark Halo A tight correlation has been reported between the mass of a central BH and the stellar velocity dispersion (σ) of the host galaxy’s bulge, as shown on Fig. 1 (Ferrarese & Merritt 2000; Gebhardt et al. 2000): (cid:8) (cid:9) σ β Mh =α 200 km s−1 M(cid:1) (3) where α = (1.7 ± 0.3) × 108, β = (4.8 ± 0.5) (Ferrarese & Merritt 2000), andα=(1.2±0.2)×108,β =(3.8±0.3)(Gebhardtetal.2000).Tremaine et al. (2002) assign the range in quoted values for β to systematic differences invelocitydispersionsusedbydifferentgroups.TheM –σ relationreported bh originally in the literature (Gebhardt et al. 2000; Ferrarese & Merritt 2000; Tremaineetal.2002)isprimarilybasedonlocalearly-typegalaxies(E/SOs) andahandfulofspiralsSb–Sbc,anditprimarilysamplesquiescentBHswith masses in the range a few × (107–109) M(cid:1). This relation was subsequently found to also hold in AGN hosts (Ferrarese et al. 2001), and in bright QSOs outtoz∼3withestimatedBHmassesofupto1010 M(cid:1) (Shieldsetal.2003). This suggests that active and quiescent BHs bear a common relationship to the surrounding triaxial component of their host galaxies over a wide range of cosmic epochs and BH masses (106–1010 M(cid:1)). NumerousvariantsoftheM –σ relationhavebeenproposed.Whileear- bh liercorrelationsbetweenthemassofCDOs/SMBHsandthebulgeluminosity (L ) had significant scatter (e.g., Kormendy & Richstone 1995), recent bulge work (Ha¨ring & Rix 2004) based on improved BH and bulge masses yield Fueling AGN and Starbursts 147 a very tight M –M relation. Graham et al. (2001) find a correlation bh bulge between the light concentration of galaxies and the mass of their SMBHs, andclaimthisrelationisastightastheM –σ relation.Groginetal.(2004) bh have searched for signs of this correlation at z ∼ 0.4–1.3 in a comparative study of structural parameters among 34000 galaxies in the GOODS fields, including 350 X-ray selected AGN hosts in the overlapping Chandra Deep Fields. Compared to the inactive galaxies, the AGN hosts have significantly enhancedconcentrationindicesthroughouttheentireredshiftrange,asmea- suredinrestframeB-bandforavolume-limitedsampletoM <−19.5(and B toL(2–8keV)>1042 fortheAGN).Finally,Ferrarese(2002)showsthatthe M –σ relationtranslatestoarelationbetweenthemassoftheBHandthat bh of the dark matter (DM) halo (M ) dm (cid:10) (cid:11) M 1.65 Mh =107 M(cid:1) 1012dMm(cid:1) (4) if one assumes that σ correlates with the circular speed V which bears an c intimate relation to the DM halo within the standard ΛCDM paradigm. A plethora of theoretical studies have explored the growth of BHs and thepossibleoriginofafundamentalM –σ relation(e.g.,Haehnelt&Kauff- bh mann 2000; Adams, Graff, & Richstone 2001; Burkert & Silk 2001; Di Mat- teo,Croft,Springel,&Hernquist2003;Bromm&Loeb2003;Wyithe&Loeb 2003;El-Zantetal.2003).AccordingtoHaehnelt&Kauffmann(2000),hier- archicalgalaxyformationmodelswherebulgesandSMBHsbothformduring major mergers produce a M –σ correlation. Star-formation (SF) regulated bh growth of BHs in protogalactic spheroids has been proposed by Burkert & Silk (2001) and Di Matteo et al. (2003). In many of these models, black hole growthstopsbecauseofthecompetitionwithSFand,inparticular,feedback, both of which determine the gas fraction available for accretion. According to Wyithe & Loeb (2003), a tight M –σ relation naturally results from hi- bh erarchical ΛCDM merging models where SMBHs in galaxy centers undergo self-regulated growth within galaxy haloes until they unbind the galactic gas thatfeedsthem.El-Zantetal.(2003)havesuggestedthattheBH–bulge–DM halo correlation can be understood within the framework of galactic struc- tures growing within flat-core, mildly triaxial haloes. 3 Central Issues in Fueling AGN and Starbursts I present here an overview of several central issues that are relevant for un- derstanding the fueling of AGN and circumnuclear starbursts. 3.1 Mass Accretion Rates For a standard BH accretion disk with an efficiency (cid:14) of conversion between matter and energy, the radiated bolometric luminosity L is related to the bol 148 S. Jogee Table 1. Typical L and M˙ for QSOs and local AGN bol bh L a Typical L Typical M˙ b bol bol bh Type of AGN (ergs s−1) (ergs s−1) (M(cid:1) yr−1) (1) (2) (3) (4) QSOs 1046–1048 1047–1048 10–100 Seyferts 1040–1045 1043–1044 10−3–10−2 LINERs 1039–1043.5 1041–1042 10−5–10−4 Notes to Table – a. The full range in bolometric luminosity (L ) for bol Seyfert and LINERS is taken from Ho, Filippenko, & Sargent 1997a, while for QSOs different sources in the literature are used; b. The typ- ical M˙ in column (4) is derived from the typical L in column (3) bh bol assuming a standard radiative efficiency (cid:13)∼0.1 mass accretion rate (M˙ ) at the last stable orbit of a BH by bh (cid:8) (cid:9) (cid:10) (cid:11) (cid:14) L M˙bh =0.15 M(cid:1) yr−1 0.1 1045 erbgols s−1 (5) Table1showstypicalobservedbolometricluminositiesandinferredmass accretionratesforQSOsandlocalAGN(Seyfert,LINERS)assumingastan- dard radiative efficiency (cid:14) ∼ 0.1. The standard value of (cid:14) ∼ 0.1 applies if the gravitational binding energy liberated by the accreting gas at the last stable orbit of the BH is radiated with an efficiency of ∼ 0.1 c2. In prac- tice, the radiative efficiency depends on the nature of the accretion disk and gas accretion flows. For instance, thin-disk accretion onto a Kerr BH can lead to a radiative efficiency (cid:14) ∼ 0.2. It has been suggested that the most luminous quasars at high redshift may have grown with (cid:14) ∼ 0.2, or alterna- tively that they have a super-Eddington luminosity (Yu & Tremaine 2002). Conversely, in certain popular models of gas accretion flows such as adi- abatic inflow-outflow solutions (ADIOS; Blandford & Begelman 1999) and convection-dominated accretion flows (CDAF: Narayan et al. 2000) only a smallfractionofthematterwhichaccretesattheouterboundaryoftheflow contributestothemassaccretionrateattheBHduetoturbulenceandstrong mass loss. This leads to an effective radiation efficiency (cid:5) 0.1 when applied tothemassaccretionrateattheouterboundaryoftheaccretionflow.Thus, within the CDAF and ADIOS paradigms, the gas inflow rates that must be supplied on scales of tens of pc may be much larger than those quoted in Table 1, even for low luminosity Seyferts. 3.2 The Angular Momentum Problem ThemostimportantchallengeinfuelingAGNistheangularmomentumprob- lem rather than the amount of fuel per se. The angular momentum per unit Fueling AGN and Starbursts 149 ISOLATED OR WEAKLY MAJOR MERGER OF INTERACTING SPIRAL TWO DISKS R~5000 L~5x1029 Early Merger Stage Gravitationaltorquefrom Hydrodynamical torque aspontaneouslyortidally (shocks) in initial collision induced large-scale bar Gravitational torque from induced large-scale bars dominate R~500 Late Merger Stage L~5x1028 Gnersatveidt antuiocnleaalrt obraqru(se)from Rapidly varying gravitational torque. Gas on crossing orbits Dynamical friction strongly shocked SF feedback Dynamical friction Nuclear spiral (shocks) SF feedback R~10 L~1026 Runaway self-gravitational instabilities Tidal disruption of clumps by BH Viscous torques Hydromagnetic wind (pc and sub-pc scale) Rlast~ 3M8x 10-5 L~ 2M x 1024 8 Fig. 2. The angular momentum problem in the fueling of AGN and starbursts: The specific angular momentum (L) of gas located at a radius (R) of severalkpcmustbereducedbymorethan104beforeitisfitforconsumptionatthe laststableorbit(R )ofaBH.Incontrast,powerfulstarburstscanbemoreeasily last triggeredviagravitationaltorqueswhichbuildlargegasdensitiesoncircumnuclear (R = 500 pc) scales. This figure schematically illustrates some mechanisms that canreduceLanddrivegasinflowonvariousspatialscalesinarelativelyquiescent galaxy (left) and in a major merger (right). R is in pc, L is in units of cm2 s−1, and a (M8×108 M(cid:1)) BH is assumed. See text for details massorspecificangularmomentumL=r×v offuelatthelaststableradius ofaBHofmass(M8×108 M(cid:1))isseveraltimes1024M8 cm2 s−1.Incontrast, matter (star or gas) rotating in a spiral or elliptical galaxy at a radius of 10 kpc has a specific angular momentum of several times 1029M cm2 s−1. This 8 isillustrated inFig.2assumingtypicalgalactic rotation velocities.Thus,the specific angular momentum of matter located at a radius of a few kpc must be 150 S. Jogee reduced by more than 104 before it is fit for consumption by a BH. Searching for mechanisms which can achieve this miraculous reduction of angular mo- mentum is one of the driving objectives of AGN research. Even at a radius of 200 pc, L is still a factor of 1000 too large, and the angular momentum barrier is a more daunting challenge than the amount of gas. For instance, in the case of a Seyfert with an accretion rate of ∼10−2 M(cid:1) yr−1 and a duty cycle of 108 years, a gas cloud of 106 M(cid:1) may provide adequate fuel. Such cloudsarecertainlycommonwithintheinner200pcradiusofspiralgalaxies, but we yet have to understand what physical processes are able to squeeze their angular momentum out by more than 99.99%. The BH is analogous to an exigent dieter who has a plentiful supply of rich food, but can only consume 99.9% fat-free items! 3.3 Dominant Fueling Mechanisms on Different Scales Gravitationaltorques,dynamicalfriction,viscoustorques,andhydrodynam- ical torques (shocks) are some of the mechanisms which remove angular mo- mentum from the dissipative gas component and channel it to small scales, thereby helping to fuel central starbursts and massive BHs. These different fueling mechanisms assume a different relative importance at different radii in a galaxy, and also, when dealing with a strongly interacting galaxy versus an isolated one. I will review these different mechanisms in detail from an observational and theoretical perspective in Sects. 4–9, but here I discuss a few key concepts and provide a schematic overview in Fig. 2. Gravitational torques operate on a timescale (t ) comparable to the gra orbital timescale and provide, therefore, the most efficient way of reducing angular momentum on large to intermediate scales (tens of kpc – a few 100 pc).Thiscanbeseenbycomparingt withthetypicaltimescalesonwhich gra dynamicalfriction(t )andviscoustorques(t )operateforacloudofmass df vis M (Table 2). Dynamical friction on a clump of mass M and speed v at a radius R operates on a timescale which is ∝ (R2 v/M lnΛ), where lnΛ is the Coulomb logarithm (Binney & Tremaine 1987). For a 107 M(cid:1) gas cloud at a kpc radius in a disk galaxy, t is an order of magnitude larger than t df gra (Table 2). However, for massive gas clumps at low radii, dynamical friction becomes increasingly important: it can drive a 108 M(cid:1) cloud from R ∼ 200 pc down to R∼ 10 pc within a few times 107 yrs (Sect. 9). Table 2. Gravitational Torques, Dynamical Friction, and Viscous Torques R (pc) M (M(cid:1)) tgra (Myr) tdf (Myr) tvisc (Myr) (1) (2) (3) (4) (5) 1000 1e7 20 1020 1000 200 1e7 4 60 – Fueling AGN and Starbursts 151 In an isolated galaxy (Fig. 2), gravitational torques are exerted by non- axisymmetricfeaturessuchaslarge-scale(Sect.6)andnuclear(Sect.7)bars. While a large-scale bar efficiently drives gas from the outer disk into the inner kpc, the bar-driven gas flow slows or even stalls as it crosses the inner Lindblad resonance (ILR) for reasons described in Sect. 6. At this stage, the gas piles up typically at a radius of several 100 pc where powerful starbursts are commonly observed (Sect. 6; Fig. 6). However, gas on these scales has a specific angular momentum that is still more than 1000 times too high for it to be digestible by a BH. If a nuclear bar (Sect. 7) is present, it can break the status quo and torque gas from the ILR region of the large-scale bar down to tens of pc. In addition, if massive gas clumps exist in the inner few 100 pc, dynamical friction can drive them down to tens of pc (Sect. 9). Finally, feedback from SF (e.g., shocks from supernovae) can remove energy and angular momentum (Sect. 9) from a small fraction of the circumnuclear gas. On scales of tens of pc, the tidal torque from the BH itself can disrupt gas clumps and stellar clusters, possibly into an accretion disk (Sect. 9). Subsequently, on pc and sub-pc scales, viscous torques and hydromagnetic outflows in AGN (Sect. 9) may become important. Simulations suggest that induced large-scale stellar bars remain the main driver of gas inflows down to scales of a few 100 pc, even in the case of inter- acting galaxies (Fig. 2), namely in many minor mergers (Subsect. 5.2) and duringtheearlystagesofmajor(1:1)andintermediatemassratio(1:3)inter- actions (Subsect. 5.1). Just like in the case of an isolated barred galaxy, gas inflows driven by an induced bar also slow down near the ILR. However, the final stages ofamajororintermediatemassratiomergerbringinverydiffer- ent elements. As violent relaxation starts, gas experiences strongly-varying gravitational torques, and if it is on interacting and crossing orbits, it also suffers strong shocks (Subsect. 5.1; Fig. 2). Thus, in the final merger stages, gas loses angular momentum and large gas inflows ((cid:6)1 M(cid:1) yr−1) down to small scales can result, provided the earlier episodes of SF have not depleted most of the circumnuclear gas already (Subsect. 5.1). 3.4 Census and Growth Epoch of BHs Table3comparestheBHmassdensity(ρbh−qso)accretedduringtheoptically brightQSOphases(z =0.2–5) tothe BH massdensity in present-day galax- ies (both active and inactive). Yu & Tremaine (2002) find ρbh−qso ∼ (2.5 ± 0.4)×105 (h0/65)2 M(cid:1) Mpc−3 usingtheextrapolatedQSOluminosityfunc- tion from the 2dF redshift survey and a radiative efficiency of 0.1. Similar values have been reported by others including Wyithe & Loeb (2003), Fer- rarese(2002b),andChokshi&Turner(1992).Thisvalueofρbh−qso isalower limit to the total BH mass density we expect to be in place by z =0.2 since it does not incorporate optically obscured QSOs and any build-up of the BH mass occurring outside the QSO phase. However, it is probably not far off, since the BH mass density from X-ray AGN counts at z > 0.2 (ρbh−xray) 152 S. Jogee Table 3. Census of BH Mass density BH Mass Density [105 M(cid:1) Mpc−3] ρbh−QSO accreted during optical QSO phase (z = 0.2–5) 2–4a,b,c,d ρbh−Xray from X-ray background (z>0.2) 2–5e,f ρbh−local in local early-type galaxies (z<0.1) 2–6a,b,g ρbh−Sy in local Seyferts <0.5c References in table – a. Yu & Tremaine 2002; b. Wyithe & Loeb 2003; c. Ferrarese 2002; d. Chokshi & Turner 1992; e. Cowie & Barger 2004; f. Fabian & Iwasawa 1999; g. Merritt & Ferrarese 2001. is estimated to be 2–5 × 105 M(cid:1) Mpc−3 (Cowie & Barger 2004; Fabian & Iwasawa 1999; Table 3). In the local Universe, the BH mass density in early-type galaxies at z < 0.1 is estimated to be (2.5 ± 0.4) × 105 (h0/65)2 M(cid:1) Mpc−3, based on the measured velocity dispersion of early-type galaxies in the Sloan Digital Sky Survey and the M –σ relation (Yu & Tremaine 2002). However, rough bh estimatesoftheBHmassdensityinlocalactiveSeyfert1and2galaxiesyield significantly lower values (Ferrarese 2002; Padovani et al. 1990; Table 3). In summary, the census of BH mass density (Table 3) suggests that ac- cretion with a standard radiation efficiency of 0.1 during the quasar era can readily account for the BH mass density found in local (z < 0.1) early-type galaxies. Only a small fraction of this local BH mass density appears to be currently active as Seyfert galaxies and the inferred mass accretion rates in such cases are typically 103 times lower than in QSOs. This suggests that there is no significant growth of BHs in the present epoch compared to the quasar era. Thus, we should bear in mind that local AGN (Seyferts) with current low levels of BH growth may well differ from luminous QSOs near z ∼2.5inoneormoreofthefollowingcharacteristics:the nature of the dom- inant fueling mechanism, the amount of cold gas reservoir, and the nature of the host galaxy. For instance, tidal interactions and minor or major mergers may have been much more important in the quasar era and early epochs of galaxy growth than they are in activating present-day Seyfert galaxies. 3.5 The Starburst–AGN Connection While I discuss the fueling of both AGN and starbursts in this review, I will not explicitly address the starburst–AGN connection. I only mention here that this connection can be circumstantial, influential, or causal. A circumstantial connection refers to the fact that starburst and AGN activity can both manifest in the same system simply because they are affected by a common element such as a rich supply of gas, or an external trigger (e.g., an interaction). Examples include the ULIRG–QSO connection (Sanders et al. 1988), evolutionary scenarios for Seyfert 2 (e.g., Storchi-Bergmann et al.