Black Hole Demography from Nearby Active Galactic Nuclei

0.76 MB·English

by  Luis C. Ho

CarnegieObservatoriesAstrophysicsSeries,Vol.1:
CoevolutionofBlackHolesandGalaxies,2004
ed.L.C.Ho(Cambridge:CambridgeUniv. Press)
Black Hole Demography from
Nearby Active Galactic Nuclei
4
0 LUISC.HO
0 TheObservatoriesoftheCarnegieInstitutionofWashington
2

n
a
J

4
2

1
Abstract
v
7
A significant fraction of local galaxiesshow evidence of nuclear activity. I argue that the
2
5 bulkofthisactivity,whileenergeticallynotremarkable,derivesfromaccretionontoacentral
1 massiveblackhole. Thestatisticsofnearbyactivegalacticnucleithusprovideaneffective
0
probeofblackholedemography. Consistentwiththepictureemergingfromdirectdynam-
4
ical studies, the local census of nuclear activity strongly suggests that most, perhaps all,
0
/ galaxieswithasignificantbulgecomponentcontainacentralmassiveblackhole.Although
h
late-typegalaxiesappeartobegenerallydeficientinnuclearblackholes,thereareimportant
p
- exceptionsto this rule. I highlighttwo examples of dwarf, late-type galaxies that contain
o
activenucleipoweredbyintermediate-massblackholes.
r
t
s
a 1.1 Introduction
v: Thesearchformassiveblackholes(BHs)hasrecentlyenjoyeddramaticprogress,
i tothepointthatthestatisticsofBHdetectionshavebeguntoyieldusefulcluesonthecon-
X
nectionbetweenBHsandtheirhostgalaxies,thecentralthemeofthisSymposium.Lestone
r
a becomescomplacent,however,weshouldrecognizethatourknowledgeofthedemograph-
ics ofBHs in nearbygalaxies—onwhich muchof the astrophysicalinferencesdepends—
remainshighly incomplete. Direct measurementsof BH masses based on resolved gas or
stellar kinematics, while increasingly robust, are still far from routine and presently are
availableonlyforalimitednumberofgalaxies(seeBarth2004andKormendy2004). Cer-
tainlynothingapproachinga“complete”sampleexistsyet. Moreimportantly,itisfarfrom
obviousthatthecurrentstatisticsareunbiased. AsdiscussedbyBarth(2004),mostnearby
galaxiespossesschaoticnuclearrotationcurvesthatdefysimpleanalysis.Stellarkinematics
provideapowerfulalternativetothegas-basedmethod,butinpracticethistechniquethusfar
hasbeenlimitedtorelativelydust-freesystemsand,forpracticalreasons,togalaxiesofrel-
ativelyhighcentralsurfacebrightness. Thelatterrestrictionselectsagainstluminous,giant
ellipticals. Lastly, currentsurveysseverelyunderrepresentdisk-dominated(Sbc and later)
galaxies,becausethebulgecomponentinthesesystemsisinconspicuousandstarformation
tendstoperturbthevelocityfieldofthegas.
Giventheabovelimitations,itwouldbeimportanttoconsideralternativeconstraintson
BHdemography.Thiscontributiondiscussestherolethatactivegalacticnuclei(AGNs)can
playinthisregard. Thecommonlyheld,butbynowwell-substantiatedpremisethatAGNs
1
L.C.Ho
derivetheirenergyoutputfromBHaccretionimpliesthatanAGNsignifiesthepresenceofa
centralBHinagalaxy.TheAGNsignatureinandofitselfprovidesnodirectinformationon
BHmasses,butAGNstatisticscaninformus,effectivelyandefficiently,somekeyaspects
ofBHdemography. Forexample,whatfractionofallgalaxiescontainBHs? DoBHsexist
preferentiallyingalaxiesofcertaintypes?Doesenvironmentmatter?Underwhatconditions
doBHslightupasAGNsandhowlongdoestheactivephaselast? Whatistheirhistoryof
massbuild-up?Theseandmanyotherrelatedissuesareinextricablylinkedwiththestatisti-
calpropertiesofAGNsasafunctionofcosmologicalepoch.Thiscontributionconcentrates
onthelocal(z≈0)AGNs;Osmer(2004)considersthehigh-redshiftpopulation.
This review is structured as follows. I begin with an overview of the basic methodol-
ogyofthe spectralclassification ofemission-linenuclei(§1.2)bydescribingthecurrently
adoptedsystem,itsphysicalmotivation,thecomplicationsofstarlightsubtraction,andsome
practicalexamples. Section1.3brieflysummarizespastandcurrentspectroscopicsurveys
and introduces the Palomar survey. The demographics of nearby AGNs is the subject of
§1.4,coveringdetectionratesbasedonopticalsurveys,detectionratesbasedonradiowork,
thedetectionofweakbroademissionlines,issuesofrobustnessandcompletelessincurrent
surveys, the local AGN luminosity function, the statistics of accretion luminosities, host
galaxypropertiesandenvironmentaleffects,andintermediate-massblackholes.Nodiscus-
siononnearbyAGNswouldbe completewithouta propertreatmentonLINERs(§1.5). I
focus on what I believe are the three most important topics, namely the current evidence
thatthemajorityofLINERsareindeedpoweredbyaccretion,AGNphotoionizationastheir
dominantexcitation mechanismand the demise of competingalternatives, and the largely
still-unresolvednatureoftheso-calledtransitionobjects. Section1.6givesasynopsisofthe
mainpoints.
1.2 SpectralClassification ofGalacticNuclei
1.2.1 PhysicalMotivation
AGNscanbeidentifiedbyavarietyofmethods. MostAGNsurveysrelyonsome
aspect of the distinctive AGN spectrum, such as the presence of strong or broad emission
lines, an unusuallyblue continuum, or strong radio or X-ray emission. While all of these
techniquesareeffective,noneisfreefromselectioneffects. TosearchforAGNsinnearby
galaxies, where the nonstellar signal of the nucleus is expected to be weak relative to the
hostgalaxy,themosteffectiveandleastbiasedmethodistoconductaspectroscopicsurvey
of a complete, optical-flux limited sample of galaxies. To be sensitive to weak emission
lines, thesurveymustbe deepand ofsufficientspectralresolution. To obtainreliable line
intensityratios,onwhichtheprincipalnuclearclassificationsarebased,thedatamusthave
accurate relative flux calibration, and one must devise a robust scheme to correct for the
starlightcontamination.Theseissuesarediscussedbelow.Butfirst,Imustcoversomebasic
materialonspectralclassification.
Themostwidelyusedsystemofspectralclassificationofemission-linenucleifollowsthe
methodoutlinedby Baldwin, Phillips, & Terlevich(1981), and later modifiedby Veilleux
& Osterbrock (1987). The basic idea is that the relative strengths of certain prominent
emissionlinescanbeusedtoprobethenebularconditionsofasource. Inthecontextofthe
presentdiscussion,themostimportantdiagnosticisthesourceofexcitation,whichbroadly
falls into two categories: stellar photoionizationor photoionizationby a centrally located,
2
L.C.Ho
spectrallyhardradiationfield,suchasthatproducedbytheaccretiondiskofamassiveBH.
ThelatterclassofsourcesaregenericallycalledAGNs,whicharemostrelevanttoissuesof
BHdemography.
How does one distinguish stellar from nonstellar photoionization? The forbiddenlines
of the doublet[O I] λλ6300, 6364 rise from collisionalexcitation of O0 by hot electrons.
SincetheionizationpotentialofO0 (13.6eV)isnearlyidenticaltothatofhydrogen,inan
ionization-boundednebula[O I]isproducedpredominantlyinthe“partiallyionizedzone,”
whereinbothneutraloxygenandfreeelectronscoexist. InadditiontoO0,theconditionsof
thepartiallyionizedzonearealsofavorableforS+ andN+,whoseionizationpotentialsare
23.3 eV and 29.6 eV, respectively. Hence, in the absence of abundance anomalies, [N II]
λλ6548, 6583and[S II] λλ6716, 6731are strong (relativeto, say, Hα) whenever[O I] is
strong,andviceversa.
Inanebulaphotoionizedbyyoung,massivestars,thepartiallyionizedzoneisverythin
becausetheionizingspectrumofOBstarscontainsfewphotonswithenergiesgreaterthan1
Rydberg. Hence,intheopticalspectraofHIIregionsandstarburstnuclei(hereinafterH II
nuclei∗)thelow-ionizationtransitions[NII],[SII],andespecially[OI]areveryweak. By
contrast,aharderradiationfield,suchasthatofanAGNpower-lawcontinuumthatextends
intotheextreme-ultraviolet(UV)andX-rays,penetratesmuchdeeperintoanopticallythick
cloud. X-ray photoionization and Auger processes release copious hot electrons in this
predominantlyneutralregion, creating an extensivepartially ionized zone. The spectra of
AGNs,therefore,exhibitrelativelystronglow-ionizationforbiddenlines.
1.2.2 SampleSpectra
ThespectrashowninFigure1.1illustratetheempiricaldistinctionbetweenAGNs
and H II nuclei. In NGC 7741, which has a well-known starburst nucleus (Weedman et
al. 1981), [O I], [N II], and [S II] are weak relative to Hα. The [O III] λλ4959, 5007
doublet is quite strong compared to [O II] λ3727 or Hβ because the metal abundance of
NGC 7741’snucleusis rather low, althoughthe ionization levelof H II nuclei can span a
widerange,dependingonmetallicity(Ho,Filippenko,&Sargent1997c).Ontheotherhand,
thelow-ionizationlinesaremarkedlystrongerintheothertwoobjectsshown,bothofwhich
qualifyas AGNs. NGC 1358 is an example of a galaxy with a “high-ionization”AGN or
“Seyfert”nucleus.NGC1052istheprototypeoftheclassknownas“low-ionizationnuclear
emission-line regions” or “LINERs.” The ionization level can be judged by the relative
strengthsoftheoxygenlines,butinpracticeismosteasilygaugedbythe[O III]/Hβ ratio.
In the commonly adopted system of Veilleux & Osterbrock (1987), the division between
SeyfertsandLINERsoccursat[O III]λ5007/Hβ =3.0. Ho,Filippenko,&Sargent(2003)
stress, however,thatthisboundaryhasnostrictphysicalsignificance. Theionizationlevel
ofthenarrow-lineregion(NLR)inlarge,homogeneoussamplesofAGNsspansawideand
apparentlycontinuousrange; contrary to the claims of some studies (e.g., Véron-Cetty &
Véron2000)thereisnoevidenceforanyclear-cuttransitionbetweenSeyfertsandLINERs
(Hoetal. 2003;Heckman2004).
Theclassificationsystemdiscussedabovemakesnoreferencetotheprofilesoftheemis-
∗ As originally defined (Weedmanet al. 1981; Balzano 1983), a starburst nucleus is one whosecurrent star
formationrateismuchhigherthanitspastaveragerate. Thisterminologypresupposesknowledgeofthestar
formationhistoryofthesystem.Sincethisinformationisusuallynotavailableforanyindividualobject,Iwill
adoptthemoregeneraldesignationof“HIInucleus.”
3
L.C.Ho
Fig.1.1. Sampleopticalspectraofthevariousclassesofemission-linenuclei. NGC1358
= Seyfert; NGC 1052 = LINER; NGC 7714 = H II. The prominent emission lines are
identified.(BasedonHoetal. 1993aandunpublisheddata.)
sion lines. Luminous AGNs such as quasars and many “classical” Seyfert galaxies ex-
hibit permitted lines with a characteristically broad component, with FWHM widths of
∼1000- 10,000kms- 1. Thiscomponentarisesfromthebroad-lineregion(BLR),whichis
thoughttobephysicallydistinctfromtheNLRresponsibleforthenarrowlines. Following
Khachikian& Weedman (1974), it is customaryto refer to Seyfertswith and without(di-
rectly)detectablebroadlinesas“type1”and“type2” sources, respectively. Asdiscussed
in § 1.4.3,this nomenclaturecan also be extendedto includeLINERs, whichalso contain
broademissionlines. Figure1.2givessomeexamples. ThespectrumofthebrightSeyfert
galaxyNGC4151isfamiliartoall: strong,broadpermittedlinessuperposedonanunam-
biguousfeatureless,nonstellarbluecontinuuum. Butthisobjectisnottypical. Evenwithin
the Seyfert class, most objects resemble NGC 5273, where the broad component is eas-
ilyvisibleonlyforHαandthefeaturelesscontinuumisheavilydilutedbythehostgalaxy
light.ThesameappliestoLINERs(e.g.,NGC3998),wherethehostgalaxydilutioniseven
moreextreme;nonetheless,withcarefulstarlightsubtraction(§1.2.4)andprofilemodeling
(§1.4.3),onecandetectbroadHαemissioninmanyLINERs.
4
L.C.Ho
Fig.1.2. Sampleopticalspectraof broad-lineAGNs. NGC4151= “classical” Seyfert1;
NGC5273=typicallow-luminositySeyfert1;NGC3998=LINER1. (BasedonHoetal.
1993aandunpublisheddata.)
1.2.3 DiagnosticDiagrams
TheclassificationsystemofVeilleux&Osterbrock(1987),whichIadoptthrough-
out this paper, is based on two-dimensional line-intensity ratios constructed from [O III]
λ5007,Hβλ4861,[OI]λ6300,Hαλ6563,[NII]λ6583,and[SII]λλ6716,6731(hereHβ
andHα refer onlyto the narrowcomponentof the line). The main virtuesof this system,
shown in Figure 1.3, are (1) that it uses relatively strong lines, (2) that the lines lie in an
easily accessible region of the optical spectrum, and (3) that the line ratios are relatively
insensitivetoreddeningcorrectionsbecauseofthecloseseparationofthelines. Thedefini-
tionsofthevariousclassesofemission-lineobjectsaregiveninHo,Filippenko,&Sargent
(1997a)∗. In addition to the three main classes discussed thus far—H II nuclei, Seyferts,
andLINERs—Ho,Filippenko,&Sargent(1993a)identifiedagroupof“transitionobjects”
whose [O I] strengths are intermediate between those of H II nuclei and LINERs. Since
theytendtoemitweaker[OI]emissionthanclassicalLINERs,previousauthorshavecalled
them“weak-[OI]LINERs”(Filippenko&Terlevich1992;Shields1992;Ho&Filippenko
1993). Ho et al. (1993a) postulated that transition objects are composite systems having
∗ Theclassificationcriteriaadoptedheredifferslightly,butnotappreciably,fromthoseproposedbyKewleyetal.
(2001)basedontheoreticalmodels.
5
L.C.Ho
bothanH II regionandaLINERcomponent;Iwillreturntothenatureofthesesourcesin
§1.5.3.
I note that my definition of LINERs differs from that originally proposed by Heckman
(1980b),whousedsolelytheoxygenlines: [O II]λ3727>[O III]λ5007and[OI]λ6300
> 0.33 [O III] λ5007. The two definitions, however, are nearly equivalent. Inspection
of the full optical spectra of Ho et al. (1993a), for example, reveals that emission-line
nucleiclassifiedasLINERsbasedontheVeilleux&Osterbrockdiagramsalmostinvariably
also satisfy Heckman’s criteria. This is a consequence of the inverse correlation between
[OIII]/Hβ and[OII]/[OIII]inphotoionizedgaswithfairlylowexcitation([OIII]/Hβ ∼<3;
seeFig. 2inBaldwinetal. 1981).
1.2.4 StarlightSubtraction
Theschemeoutlinedabove,whileconceptionallysimple,overlooksonekeyprac-
tical complication. The integrated spectra of galactic nuclei include emission from stars,
whichin most nearbysystems overwhelmsthe nebularline emission. Thiscan be seen in
Figure 1.1, or from a cursory examination of the spectral atlas of Ho, Filippenko, & Sar-
gent (1995). Any reliable measurement of the emission-line spectrum of galactic nuclei,
therefore,mustproperlyaccountforthestarlightcontamination.
An effective strategy for removing the starlight from an integrated spectrum is that of
“template subtraction,” whereby a template spectrum devoid of emission lines is suitably
scaledtoandsubtractedfromthespectrumofinteresttoyieldacontinuum-subtracted,pure
emission-linespectrum. A numberofapproacheshavebeenadoptedto constructthe tem-
plate.Theseinclude(1)usingthespectrumofanoff-nuclearpositionwithinthesamegalaxy
(e.g., Storchi-Bergmann, Baldwin, & Wilson 1993); (2) using the spectrum of a different
galaxy devoid of emission lines (e.g., Costero & Osterbrock 1977; Filippenko & Halpern
1984;Hoetal. 1993a);(3)usingaweightedlinearcombinationofthespectraofanumber
differentgalaxies,chosentobestmatchthestellarpopulationandvelocitydispersion(Hoet
al. 1997a); (4)usingthe spectrumderivedfroma principal-componentanalysisofa large
set of galaxies (Hao & Strauss 2004); and (5) using a model spectrum constructed from
populationsynthesistechniques,usingasinputalibraryofspectraofeitherindividualstars
(e.g.,Keel1983a)orstarclusters(e.g.,Bonatto,Bica,&Alloin1989;Raimannetal. 2001).
Figure1.4illustratesthestarlightsubtractionprocessfortheH II nucleusinNGC3596
andfortheSeyfert2nucleusinNGC7743,usingthemethodofHoetal. (1997a). Givena
listofinputspectraderivedfromgalaxiesdevoidofemissionlinesandaninitialguessofthe
velocitydispersion,aχ2-minimizationalgorithmsolvesforthesystemicvelocity,theline-
broadeningfunction,the relative contributionof the variousinputspectra, and the general
continuum shape. The best-fitting model is then subtracted from the original spectrum,
yieldinga pureemission-line spectrum. In the case of NGC 3596, the modelconsisted of
the combinationof the spectrum of NGC 205, a dE5 galaxywith a substantial population
of A-type stars, and NGC 4339, an E0 galaxy having a K-giant spectrum. Note that in
the original observed spectrum (top), Hγ, [O III] λλ4959, 5007, and [O I] λ6300 were
hardlyvisible,whereasafterstarlightsubtraction(bottom)theycanbeeasilymeasured.The
intensities of both Hβ and Hα have been modified substantially, and the ratio of the two
[SII]λλ6716,6731lineschanged.TheeffectivetemplateforNGC7743madeuseofNGC
205,NGC4339,andNGC628,anScgalaxywithanucleusdominatedbyAandFstars.
Some studies (e.g., Kim et al. 1995) implicitly assume that only the hydrogenBalmer
6
L.C.Ho
Fig. 1.3. Diagnostic diagrams plotting (a) log [O III] λ5007/Hβ versus log [N II]
λ6583/Hα,(b)log[OIII]λ5007/Hβversuslog[SII]λλ6716,6731/Hα,and(c)log[OIII]
λ5007/Hβ versuslog[O I] λ6300/Hα. The nuclearspectralclasses shownareH II nuclei
(asterisks), Seyfert nuclei (squares), LINERs (solid circles), and transition objects (open
circles). (AdaptedfromHoetal. 1997a.)
linesarecontaminatedbystarlight,andthattheabsorption-linecomponentcanberemoved
bysubtractingaconstantequivalentwidth(2–3Å).Thisprocedureisinadequateforanum-
ber of reasons. First, the stellar population of nearby galactic nuclei, although relatively
uniform, is by no means invariant(Ho et al. 2003). Second, the equivalentwidths of the
differentBalmerabsorptionlineswithineachgalaxyaregenerallynotconstant. Third,the
BalmerabsorptionlinesaffectnotonlythestrengthbutalsotheshapeoftheBalmeremis-
sionlines. Andfinally,astheaboveexamplesshow,starlightcontaminateslinesotherthan
justtheBalmerlines.
7
L.C.Ho
Fig. 1.4. Illustration of the method of starlight subtraction. In each panel, the top plot
showstheobservedspectrum,themiddleplotthebest-fitting“template”usedtomatchthe
stellarcomponent,andthebottomplotthedifferencebetweentheobjectspectrumandthe
template.InthecaseofNGC3596(a),themodelwasconstructedfromNGC205andNGC
4339,while forNGC7743(b), themodelwas derivedfroma linearcombinationof NGC
205,NGC4339,andNGC628.(AdaptedfromHoetal. 1997a.)
1.3 Spectroscopic SurveysofNearbyGalacticNuclei
Itwasapparentfromsomeoftheearliestredshiftsurveysthatthecentralregionsof
galaxiesoftenshowevidenceofstrongemissionlines(e.g.,Humason,Mayall,&Sandage
1956). A numberof studiesalso indicatedthatin manyinstancesthe spectrarevealedab-
normalline-intensity ratios, most notablythe unusuallygreat strengthof [N II] relative to
Hα (Burbidge & Burbidge 1962, 1965; Rubin & Ford 1971). That the optical emission-
linespectraofsomenucleishowpatternsoflowionizationwasnoticedfromtimetotime,
primarilybyOsterbrockandhis colleagues(e.g., Osterbrock& Miller 1975; Koski& Os-
8
L.C.Ho
terbrock1976;Costero&Osterbrock1977;Grandi&Osterbrock1978;Phillips1979),but
alsobyothers(e.g.,Disney&Cromwell1971;Danziger,Fosbury,&Penston1977;Fosbury
etal. 1977,1978;Penston&Fosbury1978;Stauffer&Spinrad1979).
Mostoftheactivityinthisfieldculminatedinthe1980s,beginningwiththerecognition
(Heckman,Balick,&Crane1980;Heckman1980b)ofLINERsasamajorconstituentofthe
extragalacticpopulation,andthenfollowedbyfurthersystematicstudiesoflargersamples
of galaxies (Stauffer1982a, b; Keel 1983a, b; Phillips et al. 1986; Véron & Véron-Cetty
1986;Véron-Cetty&Véron1986;seeHo1996formoredetails).Thesesurveysestablished
three importantresults: (1) a large fraction of local galaxies contain emission-line nuclei;
(2)manyofthesesourcesareLINERs;and(3)LINERsmaybeaccretion-poweredsystems.
Despitethesuccessoftheseseminalstudies,therewasroomforimprovement.Although
mostofthesurveysattemptedsomeformofstarlightsubtraction,theaccuracyofthemeth-
ods used tended to be fairly limited (see discussion in Ho et al. 1997a), the procedure
wassometimesinconsistentlyapplied,andinsomeofthesurveysstarlightsubtractionwas
largely neglected. The problem is exacerbated by the fact that the apertures used for the
observationswerequitelarge,therebyadmittinganunnecessarilylargeamountofstarlight.
Furthermore,mostofthe datawerecollectedwith ratherpoorspectralresolution(FWHM
≈10Å).Besideslosingusefulkinematicinformation,blendingbetweentheemissionand
absorptioncomponentsfurthercompromisestheabilitytoseparatethetwo.
Thus,itisclearthatmuchwouldbegainedfromasurveyhavinggreatersensitivitytothe
detectionofemissionlines.Thesensitivitycanbeimprovedinatleastfourways—bytaking
spectrawithhighersignal-to-noiseratioandspectralresolution,byusinganarrowerslitto
betterisolatethenucleus,andbyemployingmoreeffectivemethodstohandlethestarlight
correction.
ThePalomarspectroscopicsurveyofnearbygalaxies(Filippenko&Sargent1985,1986;
Ho et al. 1995, 1997a–e, 2003) was designed with these goals in mind. Using a double
CCDspectrographmountedontheHale5-mreflectoratPalomarObservatory,high-quality,
moderate-resolution, long-slit spectra were obtained for a magnitude-limited (B ≤ 12.5
T
mag)sampleof486northern(δ>0◦)galaxies.Thespectrasimultaneouslycoverthewave-
length ranges 6210–6860Å with ∼2.5 Å resolution (FWHM) and 4230–5110Å with ∼4
Åresolution. Mostoftheobservationswereobtainedwithanarrowslit(generally2′′,and
occasionally1′′), and the exposuretimes were suitably long (up to 1 hr or more for some
objectswithlowcentralsurfacebrightness)tosecuredataofhighsignal-to-noiseratio.This
surveycontainsthelargestdatabasetodateofhomogeneousandhigh-qualityopticalspec-
tra of nearby galaxies. It is also the most sensitive; the detection limit for emission lines
is∼0.25Å,roughlyanorder-of-magnitudeimprovementcomparedtopreviouswork. The
selectioncriteriaofthesurveyensurethatthesamplegivesafairrepresentationofthelocal
(z ≈ 0) galaxy population, and the proximity of the objects (median distance = 17 Mpc)
enablesrelativelygoodspatialresolutionto be achieved(typically∼<200pc). Theseprop-
erties of the Palomar survey make it ideally suited to address issues on the demographics
andphysicalpropertiesofnearby,andespeciallylow-luminosity,AGNs. Unlessotherwise
noted, the main results presented in the rest of this paper will be taken from the Palomar
survey.
9
L.C.Ho
Fig.1.5. (Left)Percentageofgalaxieswiththevariousclassesofemission-linenucleide-
tected as a function of Hubble type. (Right) Distribution of the classes of emission-line
nucleiasafunctionoftheabsoluteBmagnitudeofthehostgalaxy.Thearrowineachpanel
marksthemedianofthedistribution. Thesolidandhatchedhistogramsdenotetype1and
type1+type2sources,respectively.(AdaptedfromHoetal. 1997b.)
1.4 DemographicsofNearbyAGNs
1.4.1 DetectionRates
In qualitative agreement with previous surveys, the Palomar survey finds that a
substantialfraction(86%)ofallgalaxiescontaindetectableemission-linenuclei(Hoetal.
1997b).Thedetectionrateisessentially100%foralldisk(S0andspiral)galaxies,andover
50%forellipticalgalaxies.Oneofthemostsurprisingresultsisthelargefractionofobjects
classifiedasAGNsorAGNcandidates.SummedoverallHubbletypes,43%ofallgalaxies
thatfallinthesurveylimitscanbeconsidered“active”(Fig. 1.5).Thispercentagebecomes
even more remarkable for galaxies with an obvious bulge component, rising to ∼50%–
70%forHubbletypesE–Sbc. By contrast, thedetectionrate ofAGNsdropsdramatically
toward later Hubble types (Sc and later), which almost invariably(80%) host H II nuclei.
ThisstrongdependenceofnuclearspectralclassonHubbletypehasbeennoticedinearlier
studies(Heckman1980a;Keel1983a;Terlevich,Melnick,&Moles1987).
Among the active sources, 11% have Seyfert nuclei, at least doubling older estimates
(Stauffer 1982b; Keel 1983b; Phillips, Charles, & Baldwin 1983; Huchra & Burg 1992).
LINERsconstitutethedominantpopulationofAGNs. “Pure”LINERsarepresentin∼20%
ofallgalaxies,whereastransitionobjects,whichbyassumptionalsocontainaLINERcom-
ponent,accountforanother∼13%. Thus,ifallLINERscanberegardedasgenuineAGNs
(see § 1.5), they truly are the most populousconstituents—theymake up 1/3 of all galax-
ies and 2/3 of the AGN population (here taken to mean all objects classified as Seyferts,
LINERs,andtransitionobjects).
ThesampleofnearbyAGNsemergingfromtheSloanDigitalSkySurvey(SDSS)(Kauff-
mannetal.2003;Hao&Strauss2004;Heckman2004)farsurpassesthatofthePalomarsur-
10