Ornithological Monographs 54

Ornithological
Monographs
No. 54

PopulationDynamicsof the
California SpottedOwl (Strix occidentalis
occidentalis):A Meta-Analysis
ALAN I3.FRANKLIN,R. J. GUTII•RREZ,
JAMESD. NICHOLS,MARKE.
$EAMANS,
GARYC. WHITE,GUTHRIE$. ZIMMERMAN,JAMES
E. HINES,
THOMASE. MUNTON,WILLIAM $. LAHAYE,JENNIFER
A. 13LAKESLEY,
GEORGE
N. $TEGER,
B^RR¾
R. NOON,DANIELW. H. SH^W,JOHNJ.KEANE,
TRENT L. MCDONALD, AND SUSANBRITTING

PUBLISHED

BY

THE AMERICAN ORNITHOLOGISTS' UNION


POPULATION
CALIFORNIA

DYNAMICS

SPOTTED

OF THE
OWL

(STRIX OCCIDENTALIS OCCIDENTALIS)'
A META-ANALYSIS


ORNITHOLOGICAL

MONOGRAPHS

Editor:JohnFaaborg
224 Tucker

Hall

Divisionof BiologicalSciences
Universityof Missouri
Columbia, MO 65211
ManagingEditor:BradleyR. Plummer
Proof Editors: Mark C. Penrose,Richard D. Earles
AOU

Publications

Office

622 ScienceEngineering

Departmentof BiologicalSciences
Universityof Arkansas
Fayetteville,
Arkansas72701

Ornithological
Monographs,
publishedby the American Ornithologists'Union, has
beenestablished
for majorpaperstoo longfor inclusionin the Union'sjournal,TheAuk.
Publicationhasbeenmadepossiblethroughthe generosityof the late Mrs. Carll Tucker
and Marcia BradyTuckerFoundation,Inc.
Copiesof Ornithological
Monographs
may be orderedfrom ButeoBooks,3130 Laurel
Road,Shipman,VA 22971.Priceof OrnithologicalMonographs54:$10.00($9.00for AOU
members).
Add $4.00for handlingand shippingchargesin U.S.,and $5.00for all other
countries.Make checkspayableto ButeoBooks.
Corresponding
authorof thisissue,R. J.Gutierrez.

The costof thisOrnithological
Monographwasdefrayedby theU.S.Departmentof Agriculture,
Forest Service.

Library of CongressControlNumber 2004100002

Printedby CadmusCommunications,

Ephrata,PA 17522
Issued30 January2004
Ornithological
Monographs,
No. 54 vi + 54 pp.
Copyright¸ by the AmericanOrnithologists'Union, 2004
ISBN: 0-943610-00-1

Cover:SpottedOwl (Strixoccidentalis),
ink sketchby Viktor Bakhtin.


POPULATION

DYNAMICS

CALIFORNIA

SPOTTED

OF THE
OWL

(STRIX OCCIDENTALIS OCCIDENTALIS):
A META-ANALYSIS

BY:

ALAN B. FRANKLIN,
1'2R. J. GUTII•RREZ,

3JAMESD. NICHOLS,
4 MARKE. SEAMANS,
3
GARYC. WHITE,2 GUTHRIES. ZIMMERMAN,
3JAMESE. HINES,4 THOMASE. MUNTON,5
WILLIAM S. LAHAYE,3JENNIFER
A. BLAKESLEY•
2GEORGE
N. STEGER,
5
BARRYR. NOON,2DANIELW. H. SHAW,
5JOHNJ. KEANE,6TRENTL. MCDONALD,
7
AND SUSAN BRITTING 8

•Colorado
Cooperative
FishandWildlifeUnit, Colorado
StateUniversity,
FortCollins,Colorado
80523,USA;
2Department
ofFishery
andWildlifeBiology,
Colorado
StateUniversity,
FortCollins,Colorado
80523,USA;
3Department
ofFisheries,

Wildlife,andConservation
Biology,
University
ofMinnesota,
St.Paul,Minnesota
55108,USA;
4U.S.Geological
Survey,
Patuxent
WildlifeResearch
Center,
11510American
HollyDrive,Laurel,Maryland20708,USA,
sU.S.Department
ofAgriculture,ForestService,
PacificSouthwest
Research
Station,2081EastSierraAvenue,
Fresno,California93710,USA;
6U.S.Department
ofAgriculture,ForestService,SierraNevadaResearch
Center,PacificSouthwest
Research
Station,
2121Second
Street,SuiteA101,Davis,California95616,USA;
7WEST,Inc.,2003CentralAvenue,Cheyenne,
Wyoming82001,USA;and
8P.O.Box377,Coloma,
California

95613,USA

ORNITHOLOGICAL

MONOGRAPHS
PUBLISHED

THE AMERICAN

BY

ORNITHOLOGISTS'

WASHINGTON,
2004

NO. 54

D.C.

UNION


TABLE
ABSTRACT

STUDY

OF CONTENTS

.....................................................................................................................................................

AREAS

.............................................................................................................................................

1

5

LASSEN STUDY AREA ...................................................................................................................................

7

ELDORADO STUDY AREA .............................................................................................................................

8

SIERRA STUDY AREA .....................................................................................................................................

8

SEQUOIAAND KINGS CANYON NATIONAL PARKSSTUDYAREA ...........................................................

9

SAN BERNARDINO STUDY AREA .................................................................................................................
METHODS

.......................................................................................................................................................

FIELD METHODS ............................................................................................................................................

9
10
10

I0
Surveys....................................................................................................................................................
11
Estimation
ofreproductive
effort ............................................................................................................

I1
Capture,
banding,
sexandageidentification,
andresighting
ofowls ...............................................
11
Pre-analysis
datascreening...................................................................................................................
12
Meta-analysis
workshop
format ...........................................................................................................

DATA ANALYSIS ............................................................................................................................................

12

12
Changes
in analytical
methodology
fromprevious
Spotted
Owlstudies..........................................

13
Estimating
adultsurvival ......................................................................................................................
14
Estimating
fecundity...............................................................................................................................
16
Estimating
ratesofpopulation
change..................................................................................................

Comparison
ofSierraandSequoia
andKingsCanyon
national
parksstudyareas.........................
RESULTS

..........................................................................................................................................................

19
19

ADULT SURVIVAL ..........................................................................................................................................

19

FECUNDITY .....................................................................................................................................................

21

RATES OF POPULATION CHANGE ................................................................................................................

23

27
Meta-analysis
across
studyareas...........................................................................................................

COMPARISON
OFSIERRAAND SEQUOIAAND KINGSCANYONNATIONAL PARKSSTUDYAREAS .....
DISCUSSION

30

..................................................................................................................................................

30

GENERAL INFERENCES ..................................................................................................................................

31

31
Apparentsurvival ....................................................................................................................................
32
Fecundity................................................................................................................................................
33
Population
trend.......................................................................................................................................

STUDY-AREA-SPECIFIC INFERENCES ..........................................................................................................

35

35
Lassen
studyarea .....................................................................................................................................
36
Eldorado
studyarea .................................................................................................................................
37
SierraandSequoia
andKingsCanyonnationalparksstudyareas....................................................
39
SanBernardino
studyarea.....................................................................................................................

CONCLUSION AND RECOMMENDATIONS ...................................................................................................

ACKNOWLEDGMENTS
LITERATURE
APPENDICES

CITED

.............................................................................................................................

40
41

...................................................................................................................................

42

.................................................................................................................................................

48


From the Editor

With Ornithological
Monographs
#54,theAmericanOrnithologists'
Union implementsa new philosophyin the productionof its monographseries.Sincethe seriesbeganin 1964,Ornithological
Monographs
havebeenpublishedsporadically,primarily to presentarticlesthat were too large to
appearin TheAuk.Someof thosemonographs

havebeenenormous(over1,000pages),although
manywerein the50-100pagerange.Theyweresoldasseparateissues,with pressrunsof at most
a few thousandcopies.
Thisandsubsequent
monographs
will be providedto all AOU memberson a regularbasis,packagedwith TheAuk.Thereare many ornithologicalresearcheffortsthat theAOU and I feel deserve
to be publishedin one setting,without beingseparatedinto two or threemanuscriptsthat appear
in differentjournals.If youhavea dissertation,
majorresearchproject,or evena smallsymposium
longerthanthe50pagesallowedby TheAuk,we hopeyouwill considerpublishingin Ornithological
Monographs.
Ornithological
Monographs
is opento all aspectsof ornithology.All thatwe askis that theresearch
involvegoodscience,havereasonablybroadornithologicalinterest,and cantruly justify the need
for monographic
treatment.Financialsupportfor publicationis not a requirement,althoughit can
certainlyhelp the AOU and may be necessary
for largervolumes.
We begin the "new" Ornithological
Monographs
with an analysisof the demographyof the
CaliforniaSpottedOwl. Althoughthe SpottedOwl hasbecomethe focalspeciesfor both sidesin
argumentsaboutforestrypractices
in thewesternUnitedStates,mostof thenationalpublicityhas
involvedthe NorthernSpottedOwl of northwestern
California,Oregon,andWashington.
Similar
controversynow surroundsthe CaliforniaSpottedOwl. An attemptto haveit listedas an endangeredspeciesendedup in the courts,which forcedthe U.S. Fishand Wildlife Serviceto conducta
statusreview.The U.S. Fishand Wildlife Servicechosenot to list the owl, in part becausethe U.S

ForestServicehad developeda management
plan (theSierraFramework)designedto protectthe
owl and many other resourcesof the SierraNevada.However,on the day that the U.S. Fish and
Wildlife Service announced that it would not list the owl, the U.S. Forest Service announced its

desireto "modify"the SierraFramework.The modifiedframeworkwill be completedin early2004,
and it is unclearat thistime whetherthiswill be potentiallyharmful to the owl. We will undoubtedly hear more aboutthis situationin the future.
As with mostthreatenedor endangeredspecies,
we need soliddata on demographicpatterns
acrossthe species'range.This monographprovidessuchdata,combiningmodernmethodssuch
as meta-analysis
with sophisticated
capture-recapture
modelsacrossa varietyof Californiasites
The topicis criticalfor conservationpurposes,and the approachwill introducereadersto the stateof-the-artin the conductingdemographicstudies.With an interestingand importantbird, 16 wellqualifiedauthors,and pioneeringmethodsof analysis,we believethisstudysetsa high standard
for the new Ornithological
Monographs.
Any scientific
editorwill admitthatoutsidereviewis criticalto thescientific
publishingprocess
Findingreviewersfor the long manuscripts
that are potentialOrnithological
Monographs
will per-

hapsbe a challenge,
but we hopereaderswill be asexcitedabouttheconceptaswe areandwill be
willing to volunteertimewhennecessary.
Forthismonograph,
JeffreyR. Walters,EvanCooch,and

KennethH. Pollockof the AOU Conservation
Committeedid an exceptionally
detailedreviewof
an earlydraft.KatieDugger,JeffreyR. Walters,and a reviewerwho wishesto remainanonymous
madecommentson what becamethe final product.We thankthesereviewersfor the considerable
time theycontributedtowardmakingthisnew monographa high-quality,interesting,andimportant pieceof science.We alsowant to thank KimberlySmith,Brad Plummer,Mark Penrose,and
RichardEarlesof theAOU Publications
Officefor helpingtrainthisnew editorin theart of producing scientificpublications.
JohnFaaborg


Ornithological
Monographs
Volume (2004),No. 54, 1-54

POPULATION

DYNAMICS

OF THE CALIFORNIA

SPOTTED

OWL

(STRIX OCCIDENTALISOCCIDENTALIS):A META-ANALYSIS
ALANB. FRANKLIN,
1'2'9
R. J. GUTI12RREZ,
3'1ø

JAMES
D. NICHOLS,
4MARKE. SEAMANS,
3
GARYC. WHITE,2 GUTHRIES. ZIMMERMAN,
3JAMES
E. HINES,4THOMAS
E. MUNTON,
5
WILLIAMS. LAHAYE,
3'nJENNIFER
g. BLAKESLEY•
2GEORGE
N. STEGER•
5
BARRY
R. NOON,2DANIELW. H. SHAW,
5JOHNJ. KEANE,
6TRENTL. MCDONALD,
7
AND SUSAN BRITTING 8

•Colorado
Cooperative
FishandWildlife
Unit,Colorado
StateUniversity,
FortCollins,
Colorado
80523,USA;

2Department
ofFishery
andWildlife
Biology,
Colorado
StateUniversity,
FortCollins,
Colorado
80523,USA;
3Department
ofFisheries,
Wildlife,
andConservation
Biology,
University
ofMinnesota,
St.Paul,Minnesota
55108,USA,
4U.S.Geological
Survey,
Patuxent
Wildlife
Research
Center,
11510American
HollyDrive,Laurel,
Maryland
20708,USA,
sU.S.Department
ofAgriculture,

Forest
Service,
Pacific
Southwest
Research
Station,
2081EastSierraAvenue,
Fresno,California93710,USA;

6U.S.Department
ofAgriculture,
Forest
Service,
SierraNevada
Research
Center,
Pacific
Southwest
Research
Station,
2121Second
Street,SuiteA101,Davis,California95616,USA;
7WEST,
Inc.,2003CentralAvenue,Cheyenne,
Wyoming
82001,USA;and
8P.O.Box377,Coloma,California95613,USA

ABSTRACT.
--We conducted

a meta-anaiysis
to providea currentassessment
of thepopulation

characteristics
ofCalifornia
Spotted
Owls(Strixoccidentalis
occidentalis)
resident
onfourstudyareasin theSierraNevadaandonestudyareain southern
California.
Ourmeta-analysis
followed
rigorousa priorianalysis
protocols,
whichwe derivedthroughextensive
discussion
duringa
week-long
analysis
workshop.
Because
thereis greatinterestin theowl'spopulation
status,we
usedstate-of-the-art
analytical
methods
to obtainresultsasprecise
aspossible.

Our meta-analysis
includeddatafrom five Californiastudyareaslocatedon the Lassen
NationalForest(1990-2000),EldoradoNationalForest(1986-2000),SierraNationalForest
(1990-2000),Sequoiaand Kings Canyonnationalparks (1990-2000),and San Bernardino
NationalForest(1987-1998).
Four of the five studyareasspannedthe lengthof the Sierra
Nevada,whereasthefifthstudyareaencompassed
theSanBernardino
Mountainsin southern
California.Studyareasrangedin sizefrom343km2(Sequoia
andKingsCanyon)to 2,200km2
(Lassen).
All studiesweredesigned
to usecapture-recapture
methodsandanalysis.
We used
survivalin a meta-analysis
because
fieldmethodswerevery similaramongstudies.However,

we did not usereproduction
in a meta-analysis
because
it wasnot clearif variationamong
individualstudy-area
protocols
usedto assess
reproductive
outputof owlswouldconfound
results.Thus,we analyzedfecundityonlyby individualstudyarea.Weexamined

population
trendusingthereparameterized
Jolly-Seber
capture-recapture
estimator(At).
We did not estimatejuvenilesurvivalratesbecause
of estimation
problemsandpotential
biasbecause
of juvenileemigrationfrom studyareas.We usedmark-recapture
estimators
underaninformation
theoretic
frameworkto assess
apparentsurvivalratesof adultowls.The

pooled
estimate
foradultapparent
survival
forthefivestudyareas
was0.833,whichwaslower

thanpooled
adultsurvival
rates(0.850)
from15Northern
^Spotted
Owl(S.o.caurina)
studies.

Estimatesof survival from the best model on the Lassen( •b= 0.829,95% confidenceintervals

[CI]= 0.798to 0.857),Eldorado( •b= 0.8•15,
95%CI = 0.772to 0.851),Sierra( •b= 0.818,95%CI =
0.781to 0.850),and SanBernardino( •b= 0.813,95%C! = 0.782to 0.841)were not different.
However,theSequoia
andKingsCanyonpopulation
hada highersurvivalrate( •b= 0.877,95%
CI = 0.842to 0.905)than the otherstudyareas.Managementhistoryand foreststructure(e.g.
presence
of giantsequoia[Sequoiadendron
giganteum])
ontheSequoia
andKingsCanyonstudy
areadifferedfromall otherstudyareas.Thereappearsto belittleor no evidencefor temporal
variationin adultapparentsurvivalon anyof the studyareas.
Althoughwe didnotdirectlycomparefecundity,
estimates
werehighlyvariableamongyears
withinall studyareas(CV of temporalprocess
variation= 0.672-0.817).
Estimates
forfecundity
9E-mail: alanf@cnr. colostate.edu

høE-mail:

nPresentaddress:P.O.Box523,BigBearCity,California92314,USA.

2

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

amongthestudypopulations
wereLassen
(• = 0.336,SE= 0.083),Eldorado
(• = 0.409,SE=
0.087),Sierra( • = 0.284,SE= 0.073),Sequoia
andKingsCanyon( • = 0.289,SE= 0.074),andSan
Bernardino
( • = 0.362,SE= 0.038).Duringmostyears,theSierraNevadapopulations
showed
eithermoderateor poor fecrmdity.However,1992appearedto be an exceptionalreproductive year for owls in the SierraNevada. In contrast,the San Bernardinopopulationhad less
variablereproduction(CV of temporalprocessvariation= 0.217),but experienced
neitherthe
exceptionalreproductionof 1992nor the extremelypoor yearsthat characterizedall of the
SierraNevadas•udyareas.Becausefecrmditymay be influencedby weatherpatterns,it was
possiblethat the differentweatherpatternsbetweensouthernCaliforniaand the SierraNevada
accotinted for that difference.

Except
forEldorado,
allestimates
for_A
twere<1.0,butnonewasdifferent

fromA= 1.0given

the95%confidence
intervals
(Lassen
[)•=0.985,
SE= 0.02__6];
Eldorado
[•= 1.042,
SE=0.047];
Sierra
[• =0.961,
SE=0.024];
Sequoia
andKings
Canyon
[)•=0.984,
SE=0.047];
San
Bernardino
[)•=0.978,SE= 0?25]).However,
additional
evidence
(in theformof realizedpopulation
changebasedon )•t) stronglysuggested
that the Sierrapopulationdeclinedduringthe study
period.Estimated
trendsin At for theEldoradoandSierras•udyareaswerenegative.
Thus,we
couldnot distinguishdefinitivelybetweenalternativesthatthepopulationswerestationaryor

thattheestimates
ofAt werenotsufficiently
preciseto detectdeclines
onfourof thestudyareas
(Eldorado,Lassen,SanBernardino,and Sequoiaand KingsCanyon).
Resultsof the trendanalysesdo not allow stronginferenceaboutthe declineof the populations.Because
Atreflects
changes
in owlnumbers
onthes•udyareas(i.e.it integrates
emigration,
immigration,birth,anddeathrates),it doesnot allowinferenceaboutthelargerpopulationsin
whichthoselocalpopulationsareimbedded.Thatis, it is possiblethatlocalpopulationscould
be producingfewer youngbut are enhancedby immigrationfrom surroundingareas.It also
is possiblethat the conditionswithin the s•udyareasmay havebeenbetter,in termsof habitat
loss,thansurrorrnding
areasbecausea SpottedOwl conservation
strategywasimposedin the
nationalforests•udyareas.Thatmaybe particularlytrue wherehigh amountsof privateland
surroundedstudyareas(e.g.Lassen,Eldorado).Therelativelylow survivalratescoupledwith
trend estimatesthat were either decliningor <1.0 suggesta cautiousapproachto developing
conservation
strategies
for theCaliforniaSpottedOwl rmtilfurtheranalysescanbe conducted
thatcoupleclimaticandhabitatconditionswith populationparameters,
suchasadultsurvival
and fecundity.

RESUMEN.--Realizamos
un meta-anfilisis

para presentaruna evaluaci6nac•ualizadade las
caracterlsticas
de las poblaciones
de la lechuzaStrixocddentalis
occidentalis
residentesen cuatro
fireasde es•udioubicadasen la SierraNevada yen un firea de es•udioubicadaen el sur de
California.Nuestrometa-anfilisis
sigui6unosprotocolos
a prioririgurosos,
loscualesdesarrollamos a travis de discusJones extensivas dcrrante un taller de anflisis de una semana de duraci6n.

Debidoa queexistegraninternsen el esta•uspoblacional
de estalechuza,nosesforzamos
para
utilizarm6todosanallticos
modernosquearrojarlanresultados
cientfficamente
defendibles.
Nuestrometa-anflisisincluy6datosde cincositiosde estudiode California,localizadosen
el BosqueNacionalLassen(1990-2000),el BosqueNacionalEldorado(1986-2000),el Bosque
NacionalSierra(1990-2000),losparquesnacionales
Sequoiay KingsCanyon(1990-2000),y el
BosqueNacionalSanBernardino(1987-1998).
Cuatrode estossitiosseextendieron
a lo largode
la SierraNevada,mientrasqueel quintocomprendi6lasmontafiasde SanBernardino,
en el sur
deCalifornia.Lasfireasdees•udiovariaronenextensi6n
entre343km2(Sequoia

y KingsCanyon)
y 2,200km2(Lassen).Todoslos es•udiosfuerondisefiadospara usarm•todosy anfilisisde marcajey recap•ura.Utilizamosdatosde supervivencia
en un meta-anflisisporquelosm6todosde
Weld fueron muy similaresentre estudios.Sin embargo,no utilizamosla reproducci6nen un
meta-anflisis
porqueno fue clarosila variaci6nentrelosprotocolos
empleados
paradeterminar
el rendimientoreproductivoen las diferentesfireaspodrlaafectarlos resultados.
Por lo tanto,
analizamosla fecundidadde cadafrea de estudioseparadamente.
Examinamoslas tendencias
dela poblaci6n
usandoel estimador
re-parametrizado
de cap•ura-recaptura
deJolly-Seber
No estimamos
lastasasde supervivencia
de losjuvenilesdebidoa problemasde estimaci6n
y a sesgos
potencialmente
causados
por la emigraci6nde losjuvenilesdesdelasfireasde estudio, peroempleamosestimadores
de marcaje-recap•ura
bajoun marcote6ricode informaci6n
para determinarlastasasaparentesde supervivenciade laslechuzasadultas.E1valor estimado
de la supervivenciaaparentecombinadopara las cincofireasfue de 0.833,lo que es menor
que 0.850,el valor estimadode supervivenciade adultoscombinadopara 15 estudiossobre

POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL

3

la subespecie
delnor•te(S.o.caurina).
Lasestimaciones
de losmejores
modelosparala supe?
vivenciaenLassen( • = 0.829,inter•valo
deconfianza
del95%[CI]= 0.798a 0.857),•Eldorado
(•
= 0.815,CI = 0.772a 0.851),Sierra( • = 0.818,CI = 0.781a 0.850)y SanBernardino( • = 0.813,CI =
0.782a 0.841)no fuerondiferentes.Sinembargo,la poblaci6nde Sequoiay KingsCanyontuvo

unatasa
desupervivencia
mayor
( • =0.877,
CI=0.842
a0.905)
quelasdelasotras
•reas.
Lahistoriade manejoy la estructuradel bosque(e.g.presenciade sequoias
gigantes[Sequoiadendron
giganteum])
en Sequioay KingsCanyonfuerondiferentesde lasde losotrossitios.En ninguna
de las•reasde estudiopareceexistirevidenciaquesugierala existencia

de variaci6ntemporal
en la supervivenciaaparentede los adultos.
Aunque no comparamosla fecundidaddirectamente,los valoresestimadosfueron altamente variablesentre aftosen cadauno de los sitiosde estudio (coeficientede variaci6n [CV]

del procesotemporal= 0.672œ0.817).
Losvaloresestimados
de la fecundidad
en lasdistintas
poblaciones
fueron:Lassen
( b = 0.336,EE= 0.083),Eldorado
( • = 0.409,EE= 0.087),Sierra( • =
0.284,EE= 0.073),Sequoia
y KingsCanyon( • = 0.289,EE= 0.074)y SanBernardino
( • = 0.362,
EE = 0.038).Durantela mayorparte de los aftos,laspoblacionesde la SierraNevadaexhibieron
fecundidades
bajaso moderadas,mientrasque 1992pareci6ser un afio excepcional
para la

reproducci6n
de laslechuzasen estaregi6n.En contraste,
la reproducci6n
en la poblaci6nde
SanBernardinofue menosvariable(CV del procesotemporal= 0.217),y no present6la excepcionalreproducci6nde 1992,ni los aftosextremadamente
malosquecaracterizarona todaslas
•reas ubicadasen la SierraNevada.Debido a que la fecundidadpuedeserinfluenciadapor
los patronesclim•ticos,es posibleque las diferenciasencontradas
puedanexplicarsepor las
diferentescondicionesde clima del sur de Californiay la SierraNevada.

A excepci6nde Eldorado,todoslos valoresestimadosde At fueronmenoresa 1.0, pero
ninguno fue differentede A = 1.0 dados los intervalosde confianzadel 95% (Lassen[)• =

0.985,
EE= 0.026._],
Eldorado
[•= 1.042,
EE= 0.047],
SierLa
[• = 0.961,
EE= 0.024],
Sequoia
y
KingsCanyon [)• = 0.984,EE = 0.047],San Bernardino[)• = 0.978,EE = 0.025]).Sin embargo,
evidenciaadicionalenformadel cambiopoblacionalrealbasadoen )•t sugiri6fuertementeque

la poblaci6nde Sierradeclin6duranteel periodode estudio.Lastendencias
estimadasde At
paralas•reasde estudiode Eldoradoy Sierrafueronnegativas.Porlo tanto,no pudimossaber

deformadefinitivasilaspoblaciones
est•nestables,
o si losvaloresestimados
de/•t nofueron
lo suficientemente
precisoscomoparadetectardisminuciones
encuatrodelas•reasdeestudio
(Eldorado,Lassen,SanBernardino,y Sequoiay KingsCanyon).
Los resultadosde los analisisde las tendenciasno permitenhacerinferenciasfuertessobre

el declivede laspoblaciones.
Debidoa queA•reflejaloscambiosen el n•mero de lechuzasen
las•reasde estudio(i.e.integralastasasde inmigraci6n,emigraci6n,natalidady mortalidad),
esteparkmetrono permitehacerinferenciassobrelas poblaciones
m•s grandesen las que
est•ninmersaslas poblaciones
locales.Estoquieredecirque esposibleque las poblaciones
localesest•n produciendomuy pocascrfas,pero que est•n siendosuplementadas
mediante
inmigraci6ndesdelas •reas circundantes.Tambi•n es posibleque las condicionesdentro de
las •reas de estudiohayansidomejoresen t•rminos de p•rdida de h•bitat que las de las •reas
circundantes,debidoa queen losbosquesnacionalesde estudiose implement6una estrategia
de conservaci6n
para S. occidentalis.
Estopodriaserparticularmente
ciertoen sitiosrodeados
por gran cantidad de tierras privadas (e.g. Lassen,Eldorado). Las tasasde supervivencia
relativamentebajas,en combinaci6ncon las estimaciones
de las tendenciasque estuvieron
en decliveo fueronmenoresa 1.0, sugierenque se debeactuarcon cautelaal desarrollar
estrategiasde conservaci6npara S. o. occidentalishastaque se realicenanalisisque acoplen
condicionesclim•ticasy de h•bitat con par•metrospoblacionales,comola supervivenciay
fecundidad

de los adultos.

ThE CALIFORNIA
SPOTTED
Owl (Strix occiden- the other two subspecies,the Northern (S. o

talis occidentalis)
is one of three SpottedOwl caurina)and Mexican (S. o. lucida)spottedowls
subspecies.
It occursas a contiguouspopula- (Barrowclough
et al. 1999).Unlike the Northern
tion in the Sierra Nevada of California and as
and Mexicansubspecies,
the CaliforniaSpotted
insularpopulationsin centralcoastalCalifornia, Owl has not been listed as a threatenedspesouthernCalifornia,and BajaCaliforniaNorte, cies under the U.S. EndangeredSpeciesAct
Mexico (Gutierrez et al. 1995). The California Nevertheless, the status, trends, and basic
SpottedOwl is geneticallydifferentiatedfrom natural history (e.g. habitat selection)of the


4

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

California SpottedOwl have been the center issuesacrossthe Sierra Nevada (USDA Forest
of controversyfor more than a decade(Verner Service1998a).An accompanying
documentthat
et al. 1992b; U.S. Department of Agriculture summarizedcurrentmanagementdirectionspe[USDA], ForestServicet998a, b).
cificto eachof the high-priorityissueswas preVerner et al. (1992a[CaliforniaSpottedOwl paredby the USFS(USDAForestService1998b).
Report, hereafter CASPO]) evaluated the sta- Information provided in those documents
tus and trends of, and the state of ecological determinedthe scopeand focusof subsequent
knowledgeabout,the CaliforniaSpottedOwl. land-management
planningthat was conducted

Two fundamentalfindingsof CASPOwere un- aspart of the SierraNevadaFrameworkProject.
certaintyin populationtrendsof the owl because Those efforts culminated in a final EIS and
of the shortdurationof extantowl-demographic Record of Decision(ROD) that identified new
studies,and the probabledeclinethroughoutthe managementdirection for California Spotted
SierraNevadaof forestattributes(e.g.verylarge- Owls and the otherhigh-priorityissueson USFS
diametertrees) associatedwith SpottedOwls. lands acrossthe Sierra Nevada (USDA Forest
The CASPO

also recommended

a set of interim

Service2001a,b). The SierraNevada Framework

guidelinesto the USDA ForestService(USFS)for
the managementof SpottedOwl habitat.On the
basisof the strategyrecommendedby CASPO,
the ForestServiceimplementednew guidelines
(USDA ForestService1993), with the intention
of movingbeyondthe interim guidelineswhen
credible scientific information was gathered
from field studiesto justify a changefrom the
interim guidelines.The USFS then embarked
on a seriesof EnvironmentalImpact Statements
(EIS) (USDA ForestService1995,1996),the last
of which was unfavorably reviewed (Federal
Advisory Committee 1997). The Federal

Projectteam formally requestedthat a Spotted
Owl populationmeta-analysis

be conductedusingtheinformationon owl populationdynamics
gatheredbeforeand afterthe CASPOto address
the ongoing controversyregardingthe conservationstatusof California SpottedOwls. In
particular,theyrequestedthat the meta-analysis
examinetrendsin CaliforniaSpottedOwl populationsusingdabafrom five existingstudiesthat
had collecteddata on demographiccharacteristicsof CaliforniaSpottedOwls.
Meta-analysis
hasbeenemployedas an analytical tool to evaluatethe statusand trendsof
AdvisoryCommittee(1997)concluded,among Northern Spotted Owls since 1993 (Burnham
other things,that there was little new informa- et al. 1996, Franklin et al. 1999).Meta-analyses
tion on the owl'sbiologythat couldjustify the allow synthesisof data from independentstudchangesproposedin the EIS documents,citing ies where studiesare consideredthe sampling
in particularthe uncertaintyregardingpotential units (Wolf 1986, Hunter and Schmidt 1990).
impactof timberharveststrategieson owl habi- They can be performedon statisticscollected
tat and owl populationdynamics.Concomitant from published peer-reviewed papers (e.g.
to the EIS efforts,the Sierra Nevada Ecosystem Vanderwerf 1992) or from the raw data them-

selves(Franklin and Shenk 1995).The power
of a meta-analysisfor California SpottedOwls
is the ability to combineinformationfrom several studiesto achievegreatersamplesize and
perhaps investigatesourcesof variation and
examine potential correlationsin population
Owl, but it correctly framed issuesfacing the dynamics,otherwiseunavailablefrom a single
future of the SierraNevadaas ecosystem-wide. study.For example,Burnhamet al. (1996)used
Thus, the USFS abandoned earlier EIS efforts and raw data gathered from 13 Northern Spotted
initiateda new strategyto evaluatenot only.the Owl populationstudiesin thePacificNorthwest
an accelerating
declinein female
CaliforniaSpottedOwl but alsoother sensitive to demonstrate
wildlifespecies
andhabitatconditions

whilecon- survival over the period of study,which supsideringsuchissuesasthe threatof wildfire and portedthe inferenceof a populationdeclinein
timberharvestpractices.
In 1998,the USFSPacific NorthernSpottedOwls.
Coincidentwith the meta-analysis,a Pacific
SouthwestResearchStationpublisheda scientific review identifyingand synthesizingcurrent Northwest Forest Plan was developedfor the
knowledgeon the highest-priorityconservation Northern Spotted Owl and other old-forest

Project(1996)wasestablished,
whichattempted
to identifymultipleconcernsrelativeto conservation and managementof the Sierra Nevada
ecosystem.The report of the Sierra Nevada
Ecosystem
Projectdid not explicitlydiscussthe
ecologyor management
of theCaliforniaSpotted


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL

species(Thomaset al. 1993) at the requestof
(U.S.) PresidentW. J. Clinton. That plan proposeda conservationreservedesignto protect
SpottedOwl habitat,which wasbasedon earlier
conservationstrategiesfor the Northern Spotted
Owl (Thomaset al. 1990,U.S.Departmentof the
Interior [USDI] 1992). That reservedesign effectivelyremoved large areasfrom timber har-

5

al. (1999). That rigorous processwas critical to
the success

of the coordinatedanalysisand was
greatly facilitatedby the presenceand interaction of that diversegroup.
Our intent in analyzingthe datawas to examine trendsin demographicparametersaswell as
ratesof populationchange(•,) becausechanges
in demographicparameters(the integral comvest consideration, much of which was within
ponentsof •,) can provide betterunderstanding
existingSpottedOwl demographicstudy areas. of populationdynamics.For example,Franklin
Essentially,the reservedesignprovided an ad et al. (2000) suggestedthat adult survival dehoctest of the effectof habitat losson Spotted fined the magnitudeof •, whereasreproduction
Owl population trends. In 1998, a second and recruitment determined variation in •, over
meta-analysison the Northern Spotted Owl time. Thus, we did not want to rely solely on
population data was conducted (Franklin et a single measureof population trend. We did
al. 1999).In thoseanalyses,decliningtrendsin not explicitly evaluate changesin population
owl populationsand adult femalesurvivalwere numbers.Although data were available to eseither reduced or stabilized. Further, in the 1998 timate numbersof owls on severalof the study
meta-analysis,a new analytical method was areas, there were potential biases(see review
introduced: direct estimation of •, (the annual in Pollocket al. 1990) in estimatingnumbers
finite rate of populationchange,referredto here of owls using capture-recapture.Rather, we
as•) thatwasbasedon capture-recapture
data relied on the reverse-timeJolly-Seberestimaand reflectedchangesin numbersof territorial tor (hereinreferredto as •'t) to estimateanowls on study areas.That analysis(discussed nual changesin numbersof owls on the study
below) avoidedthe effectof potentiallybiased areas (Pradel 1996). We were then able to reestimatesof juvenile survival that caused un- expressthoseestimatesasrealized proportional
certainty in estimating•, with the Leslie pro- changesin numbersof owls without having to
jection matrix used in previous Spotted Owl rely on estimationof populationabundance.
studies.However, a stationarypopulation (i.e.
STUDY AREAS
•, = 1.0) using that newer analysisstill could
not demonstratedemographicstabilitybecause
stationary populations could be maintained
Demographicdata from five studyareaswere
solelyby immigrationfrom other populations. usedin the analyses(Table1, Fig. 1). SpecificatNevertheless,it was a clear attempt to incor- tributesof eachstudy area are describedin the
porate the most modern populationanalytical following sections,presentedin latitudinal ormethodsin the meta-analysis.
der from north to south.Most of the study areas

The purposeof our paper is to presentthe were considered,or included, density study
resultsof a meta-analysis
conductedusingdata areas,which were geographicallydefined areas
generated from five California Spotted Owl that weresurveyedentirelyfor SpottedOwls.
populationstudies(four within the contiguous
The Sierra Nevada was the dominant physiowl range of the Sierra Nevada and one from cal featureinfluencingthe climateon four of the
an insular population in southern California) study areas.That mountainrangehad cold,wet
to assess status and trends of some .California
winters and hot, dry summers.Winter Pacific
Spotted Owl populations. In our study, owl stormsystemswere the main sourceof precipiresearchers
from the five demographystudies, tation for the range.Thosestormsystemscould
timber industry consultants,and stakeholders be either cold or warm dependingupon their
met with expertsin analysisof populationdy- origin (e.g.Gulf of Alaska or tropicalPacific,renamicsfrom9 to 13July2001at ColoradoState spectively).SierraNevada vegetationwasheavUniversity in Fort Collins,Coloradoto conduct ily influencedby climate,elevation,aspect,and
a formal meta-analysisof all known California edaphic conditions(seebelow), which resulted
Spotted Owl population data. Participants in diverseforesttypes.
agreedto adhereto a rigid and formal protocol
Fire has been a primary force shaping the
for analyticalsessions
proposedby Andersonet distribution and structureof vegetationin the


6

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

L•ss•n

2O

0

20

40

60

Kilometers

Eldorado

20

0

20 40

60

Kilometers

Sierra

Sequoiaand

Kings

Canyon
20

0

20 40 60

Kilometers

San Bernardino

N
20

0

20 40 60

Kilometers

Fic. 1.Relativelocationsof CaliforniaSpottedOwl demographystudiesin relationto forestedhabitat(shaded
gray)throughoutCalifornia.Cross-hatched
areawithinthe Lasseninsetwasusedto estimate)•cDark-colored
circleswithin theEldoradoareowl sitesexternalto the densitystudyarea(darkshadedareawithin inset).


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL

7

Sierra Nevada becausenatural fire regimes
were characterizedby relatively frequent fire
return intervals(Skinnerand Chang 1996).Fire
return intervalsand fire behaviorhave changed
as a result of governmentalfire-suppression
policies and other vegetation-management
activities that followed European settlement
(McKelvey et al. 1996). Sierranvegetationalso
hasbeenaffectedby loggingandlivestockgrazing (McKelvey and Johnston1992). Logging
began in the mid- to late 1800s and harvest
techniquesvaried from clear felling to individual tree selection.Livestockgrazing was in-

tenseduring the 1800sbut hasbeenreducedto
relativelylow levelstoday.Theneteffectof both
natural and anthropogenicdisturbancehas led
to a complexmosaicof vegetationtypes,seral
stages,and stand structures.This has resulted
in a varietyof foresttypesusedby SpottedOwls
in the Sierra Nevada.

Each of the Sierra Nevada

demographystudy areashas slightly different
historiesof land use and vegetationhistories,
which may have influencedthe populationdynamicsof SpottedOwls (seebelow).
In contrastto the Sierra Nevada study areas,the San BernardinoMountainsstudy area
was locatedin a relatively isolatedmountain
range in southern California (Fig. 1). Owls occupying that mountain range were the largest
population of a presumed owl metapopula-

tion found throughoutthe disjunctrangesof
the region (Noon and McKelvey 1992, LaHaye
et al. 1994). The climatic environment in
southern California was more benign than the
Sierra Nevada becausethe majority of winter

stormspass to the north of the region (Karhl
1979).Loggingoccurredin the SanBernardino
Mountains from the late 1800s through the
mid-1980s (Robinson 1989, McKelvey and
Johnston1992). Commerciallogging occurred
infrequently (McKelvey and Johnston1992)
The historicfire regime in the San Bernardino
Mountainsincludedfrequentlow-intensityfires

that playeda majorrole in shapingvegetation
mosaics(Minnich 1988).However, modern fire
suppressionand historiclogginghave resulted
in significantchangesin vegetation structure
and compositionin the wetter portionsof the
mountain range (Minnich et al. 1995). Mining,
urban expansion,and numerousother human
activities also have impacted owl habitat to
somedegree(Verneret al. 1992b).


8

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

LASSEN STUDY AREA

privateland,accounting
for 63%and37%of the
EDSA,respectively.
The Lassen study area (LAS) was located
Webegandemographic
researchontheEDSA
in northeasternCalifornia, primarily on the in 1986. We initiated the RSA in 1997 to encomLassen National Forest (LNF). The greater passmoreowl territoriesandto locateowlsthat
study area encompassed
2,200 km2 and was may have emigratedfrom the EDSA. All EDSA
analogousto the RegionalStudy Area of the owl territories were on the Eldorado National
Eldorado study (see below). A subsetof LAS Forest.Thirty-eightpercentof the RSA territories were located on the Eldorado National
(-1,270 km2) was selectedfor estimationof
(seebelow) during the meta-analysis,
basedon Forest, 38% on the Tahoe National Forest, and
portionsof thestudyareasurveyedconsistently 24% on the Tahoe Basin ManagementArea.
during1992-2000.Mostprivatelandwithin the All RSA territorieswere on public land. We
study area boundarieswas not surveyed,al- did not use owls banded in the Tahoe Basin
thoughseveralowl siteson privatetimberland ManagementArea for this analysis.
adjacentto LNF were included.In addition, a
The study area was typical of the midfew sitesoverlappedLassenVolcanicNational elevation Sierra Nevada with mountainous terPark, the Plumas National Forest, and Bureau of rainbisected
bysteepriverdrainages.
Elevations
rangedfrom 366 to 2,401m. From 1962to 1995,
Land Managementland.

Elevationson the study area ranged from averageannual precipitationat the Blodgett
1,200to 2,100m. AnnuaI precipitationat 1,250- ExperimentalForest(part of the RSA; 1,340m
1,500 m averaged 141 cm in the west, 86 cm elevation)was 158 cm (Olson and Helms 1996).
in the center,and 36 cm just eastof the study Thirty-fivepercentof precipitationfell assnow,
area. Most of the precipitationfell as snow averaging 254 cm year-L Average minimum
from November through April. Averagehigh temperaturein January was IøC and average
temperaturesat the centerof the study area maximumtemperaturein Julywas28øC.
(1,380m) ranged from 6øC in Januaryto 29øC
The EDSA and RSA were typical of Sierran
•n July.Averagelow temperatures
rangedfrom Montane Forest (SMF; Kfichler 1977). From 600
-7øC in Januaryto 7øCin July.
to 1,500m, theSMFwasdominated
by ponderoMajorityof foresttypesonthestudyareawere sapineon morexericsitesandwhite fir on more
mixed conifer, with additional stands classified

mesic sites.A transition zone above 1,500 m was

astruefir. Dominanttreespecies
includedwhite dominatedby red fir (Rundelet al. 1977).Other
fir (Abiesconcolor),
sugarpine(Pinuslambertiana), commontree speciesthat occurredwithin the
ponderosapine (P. ponderosa),
incensecedar study area included sugar pine, Douglas-fir,
(Calocedrus
decurrens),
Douglas-fir (Pseudotsugaincensecedar,canyonlive oak (Q. chrysolepis),
menziesii),
red fir (A. magnifica),
andJeffreypine California black oak, Pacificdogwood (Cornus

(P.jeffreyi).Californiablackoak(Quercus
kelloggii) nuttallii),and tan oak (Lithocarpus
densifiorus).
waspresentin theunderstoryin somestands.
SIERRA STUDY AREA
ELDORADO STUDY AREA

The Sierra study area (SIE) was located
The Eldoradostudy area (ELD) consistedof -83 km east of Fresno, California in the southa 355 km2EldoradoDensityStudyArea (EDSA) ern Sierra Nevada within the watersheds of
and a 570 km2 RegionalStudy Area (RSA) lo- the San JoaquinRiver and the North Fork of
catedbetweenGeorgetownand Lake Tahoein the KingsRiver.Studywas initiatedin 1990on
the central Sierra Nevada, E1Dorado and Placer
counties, California. Boundaries for the EDSA

419km2and thenexpandedin 1994to 693km2.

were definedby the RubiconRiver, SouthFork

the landswithin the study area.

of the Rubicon River, and Middle Fork of the
American River to the south, north, and west;

The SIE was mountainouswith steepdrainagesand elevationsrangedfrom 304 m at the

The Sierra National

Forest administered

92% of

ChipmunkRidgeand BunkerHill to the north southwestern corner to 2,924 m on the eastern
and east; and Forest Road 33 to the east. The edge. Boundariesof the SIE were defined by
ELD was characterizedby a "checkerboard" USFS administrativeunits and major topodistributionof alternatingpublic (USFS)and graphicfeaturessuchas ridgesand drainages.


POPULATION DYNAMICS

OF THE CALIFORNIA

SPOTTED OWL

Annual precipitationfrom 1961 to 1990 averaged94 cmat HuntingtonLake,-16 km north of
the study areaat 2,139m in elevation(National
Oceanic and Atmospheric Administration
[NOAA] 1998). Most precipitation occurred
duringthewinter and fell mainlyassnowabove
1,220m. Summertemperaturesaveraged-16øC

9

at Huntington Lake (NOAA 1998) but could be

southernedgeof SKCand 105cmat GrantGrove
(2,013m) near the northernborder of the study
area (NOAA 1999).During winter,precipitation
fell primarily as snow above 1,220m. Average
daily temperatures
for Julyat Ash Mountainand
GrantGrovewere28 and 17øC,respectively.

Several vegetationtypes (Verner and Boss
1980)were presenton the study area in three

>38øC at lower elevations.

distinct

zones.

Low-elevation

oak woodlands

The SIE had three generalvegetationtypes: (24%of SKC below 1,220m elevation)included
oak woodlands, mid-elevation mixed conifer low-elevationpine-oak woodlands,blue oak
forests, and high-elevation conifer forests. savannas,and dense riparian deciduousforOak-woodland zone, at the lowest elevation ests.Tree speciesincludedblue oak, gray pine,
(304-1,220m), encompassed
26% of the study interior live oak, canyonlive oak, California
area and was dominatedby blue oak (Q. doug- sycamore(Platanusracemosa),
Californiabucklasii), interior live oak (Q. wislizenii),canyon eye (Aesculus
californica),
and Fremontcottonlive oak, and gray pine (P. sabiniana).
Various wood (Populus
fremontii).Large areasadjacent
foothill chaparralspecieswere abundant.Mid- to low-elevation oak woodland consisted of
elevation mixed conifer forest (1,220-2,438 m) chaparral(primarily chamise[Adenostoma
fasoccupied61% of the studyareaand was domi- ciculatum]).Mid-elevation conifer forests (67%
nated by ponderosapine, white fir, incense of SKC; 1,220-2,440 m elevation) included a
cedar,Californiablack oak, Jeffreypine, red fir, ponderosapine type at lower elevations,a midand sugarpine. A small (2 km2) groveof giant elevationripariandeciduoustypethat occurred
sequoia(Sequoiadendron

giganteum)
was within throughoutthe zone,and a mixed conifertype
that forest.High-elevationconiferforest(2,439- that wasdominantin that zone.Treespeciesin2,924m) covered13%of the studyarea and was cludedponderosapine,Jeffreypine,sugarpine,
dominatedby red fir, lodgepolepine (P.contor- white fir, red fir, incense cedar, and California
black oak. Mixed conifer forests included
10
ta), and westernwhite pine (P.monticola).
giant sequoiagroves.Basedon areasestimated
SEQUOIAAND KINGS CANYON NATIONAL PARKS
from Parsons (1994), those sequoia groves
STUDY AREA
covered7% of the study area. Sequoiagroves
were mixed conifer foreststhat containedgiThe Sequoia and Kings Canyon national ant sequoiatreesand otherconiferspecies(e.g.
parks study area (SKC) was 35 km northeast white fir and sugar pine), which were often
of Visalia and 19 km southeast of the SNF and
more numerous(Rundel1971).High-elevation
covered 343 km 2 in Fresno and Tulare counties,

coniferousforests (9% of the SKC; above 2,440

California.The SKC was managedprimarily
by Sequoiaand Kings Canyon national parks,
but the study area also included the adjacent
WhitakerForest(1.3km2)that was managedby
the Universityof California.Most of the study
area was part of the Kaweah River watershed
(primarily the North, Marble, and Middle
forks);however,the northern14 km2 was part
of the Kings River watershed.The terrain was
mountainouswith steep drainages;elevations

ranged from 427 to 3,050 m. Boundariesof the
study area were defined by U.S. Park Service
administrative boundaries and topography
(ridges and elevation). Demographic studies

m elevation)consistedprimarily of a red fir type
and a lodgepolepine type. Treesincludedred
fir, lodgepolepine,and westernwhite pine.

were initiated

on SKC in 1990.

From 1961to 1990,annual precipitationaveraged66 cm on Ash Mountain (521 m) at the

SAN BERNARDINO STUDY AREA

The San Bernardino study area (SAB) was
located in the San Bernardino

Mountains,

-140 km eastof LosAngeles,California(Fig. 1).
In 1987,the535km2BigBearStudyArea (BBSA)
wasestablished,
centeredon the majorityof the
SpottedOwl locationsknown at that time. In
1989,the size of the study area was expanded
to encompass
all foresthabitatwithin the entire

mountainrange(2,140km2).TheSanBernardino
Mountains

were

one of a series of mountain

rangesthat rise aboveextensivedesert(Vasek


ORNITHOLOGICAL MONOGRAPHS NO. 54

and Barbour 1988) and semidesert(Mooney presentan explicitdescriptionof departuresfromthe
(seealsoAppendix1).
1988) vegetationtypes in southernCalifornia standardtechniques
Surveys.-Spotted Owls were surveyed accord(Noon and McKelvey 1992,LaHaye et al. 1994).
Forestsin thosemountain rangeswere isolates ing to the methodsof Forsman(1983)and Franklin

becausetheyoccurredat elevations
higherthan
surroundingdesert and chaparral vegetation
(Noon and McKelvey 1992,LaHaye et al. 1994).
They occupied-2% of the southernCalifornia
landscape(Scottet al. 1993).
Elevationson the studyarearangedfrom 800
to 3,500 m. Climate was Mediterranean with

most precipitationfalling during the winter
months(Fujiokaet al. 1998).Annual precipitation ranged from 50 to 100 cm dependingon
location,elevation,and topography(Mirreich

1988). Those mountain ranges were mesic
comparedto the surroundinglowlands,which
allowed them to supporta diverseassemblage
of shruband forestvegetationtypes(Minnich
1998).Vegetationtypes most commonlyused
by SpottedOwls in southernCalifornia were
mixedevergreen(Sawyeret al. 1988)and montane forests (Thorne 1988). Mixed evergreen
forests occurred below -1,600

m elevation

and the dominant tree specieswere canyon

et al. (1996a).Surveyswere performedby imitating
eitherSpottedOwl territorialvocalizations
or playing
prerecordedcallsfrom a tape player.We employed
threetypesof survey:point, cruise,and walk-in surveys.We conductedthosesurveysfrom 1 April to 31
August,1986-2000,exceptin the SIE and SKC study
areas where surveyswere initiated on 1 March and

endedon 30 Septemberand the SABwhere surveys
endedon 30 September.
Night surveysconsistedof
calling at points (locations)for 10 min (15 min on
SAB)to determineif SpottedOwls were presentin a
new orhistoricallyusedarea.At eachsurveypoint,researchersimitated SpottedOwl vocalizationsto elicit
a response.We strategicallyplacedsurveypointsto
obtaincompletesurveycoverageof individualstudy
areas.Night surveyswere generallyconductedfrom

duskto 2400h. If an owl wasdetectedduringa night
survey,we conducteda walk-in survey in the same
generallocationto find its roostsand nest(if nesting),
locatea possible
mate,assess
reproductive
status,and
identifyindividuals.Weattemptedto resightall owls
locatedduringeachwalk-inand captureandbandall
unbandedowls (seebelow). We conductedadditional

walk-in surveysto captureand mark unbandedowls

live oak and big-coneDouglas-fir(Pseudotsugaif thosebirdswerenot capturedin the initial survey.
macrocarpa).
Other tree speciesassociatedwith We locatedowls during walk-in surveysby imitating
those lower

elevation

sites included

Coulter

SpottedOwl vocalizations
to elicita response
and by

pine (Pinuscoulteri),white alder (Alnusrhom- visuallysearchingthe area where the owl was debifolia),Californiasycamore,
andbig-leafmaple tectedduringthe previousnightsurvey.If no Spotted

(Acermacrophyllum).
Montaneforestsoccurred Owls were detectedduring a walk-in survey,it was
above1,600m elevationand were dominatedby termed a cruise survey. Walk-in or cruise surveys
Jeffreypine and white fir. Other treespeciesoc- lasteduntil the objectiveswere met (i.e. reproductive
and identityof individualswere determined)
curringin montaneforestsincludedsugarpine, status
or the observers deemed that further effort would
incensecedar,Californiablack oak, ponderosa
not helpaccomplish
the objectives
(e.g.an owl could
pine, western juniper (Juniperusoccidentalis),not be located within the first few hours of survey).
pinyonpine (P.monophylla),
and limberpine (P. Therefore,surveyeffortwas a functionof the actual
flexilis).
time allottedto surveysof varioustypes.
METHODS

FIELD ME2•-I ODS

MethodsamongSpottedOwl populationdynamics studies have been similar for some time; that is

We performedmultiplecompletesurveysof each
entirestudyareaduringthe courseof a field season.
In a studyof NorthernSpottedOwls,Reidet al. (1999)
detectedall eight radiomarkednonjuvenilemalesin
their studywithin three10-minvocalsurveys,which
were spacedone week apart. All studiesreported
herein conducted_>3surveysat multiple point loca-

particularlytrue of Northern SpottedOwl studies tions within owl territories. In addition, if an owl was
to
(Forsman1983,Franklinet al. 1996a).However,there detectedat night,a walk-insurveywasconducted
is somevariationamongstudiesof CaliforniaSpotted locateanymates.Thus,surveyeffortshouldhavebeen
sufficientto detectnearlyall territorialSpottedOwls
Owls because of local differences in owl behavior, difonthe studyareas.Withineachstudyareaboundary,
ferent environmental conditions, and different initial
of land ownershipor
study objectivesthat requiredmodificationsof stan- we surveyedall areasregardless
dard protocolsused in Northern SpottedOwl stud- habitatspresent,with theexceptionofthe greaterLAS
ies. Therefore,we presenthere the most consistent where only known SpottedOwl habitator previously
methodologyused amongthe studies,but we also occupiedhabitatswere surveyed,and the RSAof the


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL
ELD where

owl territories

were selected from historic

locationsprovidedby the USFSor from territories
thatwe locatedduringsurveysconductedin 1997.We
surveyedterritoriesonthegreaterLASandRSAof the
ELD individually.Because
thoseindividualterritories
were not within the boundeddensitystudy areas,we

did not use themin estimation
of Xt (seebelowfor

assumptions
of Xt)'Priorto 1990on the ELD, funding
was not sufficientto adequatelysurvey the density
study area; thus, abundanceestimatesbefore 1990
were not comparableto estimatesfrom later years

andwerenotusedin estimation
of•'t'In SIEandSKC,
studyareaswere divided into sitesof the approximate
size of owl territories.We included only thosesites
that were surveyedconsistentlyeach year in that
analysis.In 1990,surveyefforton SKCwaslessthan
during the following years and thosedata were not
includedin our analysis(Table1). Finally,in the SAB,
we only surveyedforest habitat becauseowls only
occupiedforestedhabitatin that mountainrangeand
there were extensiveareasof nonforesthabitat(e.g.
chaparral)throughoutthe range.
Estimation
ofreproductive
effort.--Weestimatedowl
reproductiveactivity by feeding live mice to owls
duringwalk-insurveys(Forsman1983,Franklinet al.
1996a).Reproducingowls usually take offeredprey

to theirnestor young,whereasnonreproducing
owls
usuallyeat or cachethe mice.We estimatedfecundity
(i.e. the numberof femaleyoungfledgedper female;
Caughley1977)from numberof femalescheckedfor

reproductivestatusand numberof youngobserved.
We assumeda 1:1 sexratio of juvenilesfor fecundity
estimates(Steger1995,Franklin et al. 1996b).
Criteriafor inferringnonreproduction
for a pair of
owlsvariedslightlyamongstudyareas(Appendix1).
However,we useda datascreeningprocessto evaluatethe internalconsistency
of datacollectedgiventhe
methodsusedby an individual study (seebelow). In
addition,we discussed
theefficacyof eachstudyarea's
criteriaat lengthduringthe workshopto assess
if the
data were sufficientlyconsistent
and rigorousto use
in a collectivemeta-analysis.Researchers
generally
agreedtheir data provided unbiasedestimatesof fecunditywithin theirrespective
studyareas.However,
we felt that, becausedifferentprotocolswere used,it
wasnot appropriateto analyzethe studyareasjointly
without further investigation,such as comparing
estimatedfecundityfor eachstudy using protocols
employedby the other studies.Such an analysis
would haverequiredwriting programsto subsample
completedatasetsof surveydata(eachstudyareahas
conductedthousandsof surveysand eachstudy area
storesthoserecords-differently),which would have
takenmoretime than allottedfor the workshop.
Capture,banding,sexand ageidentification,

and resightingof owls.--Weattemptedto captureand band
all detectedSpottedOwls following the methods
of Forsman (1983) and Franklin et al. (1996a). We

11

capturedmost owls with noosepoles,snarepoles,
or mist nets. Once captured,we fitted all owls with
a lockingfederal(U.S.Fishand WildlifeService)aluminum band on the tarso-metatarsus
of one leg. On
the oppositeleg,we markedadultandsubadultowls
with a unique combinationof colorband and color
tab (Forsmanet al. 1996),whereaswe fitted juvenile
owls with a plasticband having a color unique to
their cohort.We refittedjuvenileswith unique color
bandsand tabswhen recapturedas territory holders
in later years.
We determinedsex of nonjuvenileowls by their
callsand behavior.Males have a lower-pitchedcall
than femalesand only femaleswere known to incubate or brood young(Forsmanet al. 1984).We did
not sexjuveniles,excepton the SIE and SKC (Steger
1995).We identifiedfour age-classes
on the basisof
plumage characteristics
(Forsman1981, Moen et al.
1991):juvenile;oneyear old (first-yearsubadult);two
yearsold (second-year
subadult);and three or more
yearsold (adult).
After initial capture,we identifiedadult and subadult owls as individualsby resightingtheir unique

colorbandsand tabs.We resightedband colorsusing binoculars.When possible,two biologistsmade
independentobservationsof the same bird's color
band-tab combination.We recapturedbirds and replacedbandswhena colortabbecamefrayedthrough
wear.When colorbandswere changed,we recorded
the

metal

band

number.

Band

loss was

minimal

(Forsmanet al. 1996).

Pre-analysis
datascreening.--Our
basicphilosophy
and framework for the meta-analysisworkshopfollowedAndersonet al. (1999).All groupsconducting
California SpottedOwl demographicresearchand
experts in demographicanalysis and parameter
estimationwere invited to attendthe meta-analysis
workshop.Further, representativesfrom the timber
industry and environmentalgroupswere invited to
attend.To our knowledge,the data analyzed representedthe extent of current data on population dynamicsof CaliforniaSpottedOwls. The demographic

parametersof interestfor the meta-analysiswere sexspecificsurvival, female fecundity,and population
rate of change.Therefore,in the interest of data
consistency,
researchers
from eachstudy area were
requestedby the organizersto summarizetheir data
in (1) a data file with a capture-historymatrix that
describedthe capture-recapturehistory of each individualowl, its federalband number,its ageat first
capture (juvenile, first-year subadult, second-year
subadult, or adult), and its sex; (2) a data file with
annual number of young fledged(0, 1, 2, or 3) for
individualterritorialowls,theirterritory,socialstatus
(pairedor singleowl), ageof the male,and ageof the
female;and (3) a datafile with a capture-history
matrix that documentedthe capture-recapturehistoryof
all individualsencounteredas territoryholders(i.e. if


12

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

an individualwasfirst bandedas a juvenileonly the
territorialportionof thehistorywasincluded),its age
at first capture,its sex,and its federal band number.

viewed (see Anderson et al. 1999). All researchers
signedthe certification.
Meta-analysisworkshop
format.--We devoted the

The latter database was created after the survival da-

firstday-and-a-halfof theworkshopto a discussion
of
themethodsusedto inferreproductiveoutputof owls
becausemethodsvariedsomewhatbetweenstudyareas (Appendix 1, seeabove).Consensus
was reached
among researchersthat the methods used by the
respectivestudies,despitetheir differences,
were appropriategiventhe studyobjectives,
location,andbehaviorof owlswithin the studyarea.In addition,we
engagedin considerablediscussionand debateabout
the natureof the analysesto be performed,appropriate inferences
to be drawngivena particularanalysis,
and advantagesand disadvantages
of differentapproaches.
We reachedagreementon the structureand
natureof analysesandwho wouldperforma specific
analysis.Our discussion
alsoled to a departurefrom
pastapproaches
for SpottedOwl populationanalyses
(see discussionof X estimationbelow). Becausewe
knew our effortwould be closelyfollowedby many
interestedparties,we developedand recordeda protocolduringtheworkshop(Appendix2).

We devotedthe remainderof the workshopto selectionof relevanta priorimodelsfor reproduction,
recapture, and survivalmodeling(seeAppendix2); thento
executing
the a priorimodels.Because
researchers
were
atdifferentstages
oftheirstudies,
we agreedthatcovariates(e.g.predpitation,
habitat)wouldnotbeincludedin
themodelingprocess.
However,we alsoagreedunanimouslythatthatwasa worthwhileendeavorto be pursuedin a futuremeta-analysis
(seebelow).Researchers
wereconvened
asa groupto discuss
particularissues,
as
they arose,that might affectthe analysisor to maintain
consistency
in the analyticalprocess.
Thus,we agreed
that the resultsof our analysiswould be a first step
in settingthe basisfor subsequent
workshops,
which
would allowmoreinclusiveanalyses.

tabasehad been checkedfor errors (see below) and

wasusedto estimatepopulationrateof change.

Althoughmost studiesdid not use DNA teststo
ascertainthe sex of juveniles,we assumeda 1:1 sex
ratio at fledgingin eachyear for the capture-history
matrix (Steger1995,Franklin et al. 1996b).We only
usedowl reproductivedata for eachstudy areathat
was consistentwith the protocolsdevelopedfor each
study(seeAppendix1). However,numberof young
found in eachterritory eachyear had to be basedon
a minimum of two visitswithin a year. An exception
couldbe madeif statistical
justificationwasprovided
that indicatedsinglevisitshad high accuracy(>85%)
for countingyoung.Regardless,
we did not compare
estimatesof fecundityacrossstudyareasin a metaanalysisbecauseof differencesin protocolused by
eachstudy.
Most researchershad to modify their database
structureto conformto the meta-analysisspecificationsfor a compatibledatabasestructure(seeabove).
There alsowas somevariationin researchprotocols
amongthe studyareas.Therefore,prior to attending
theone-weekworkshopandconducting
dataanalysis,
all researchgroupsagreedto undergo a formal data
screeningprocessto ensurequality control,to ensure
that the originalfield data matchedthe data in the
computerfiles,and to ensurethat the specificcriteria
used by a study was actuallyfollowed by that study
(i.e. datacollectionwasinternallyconsistent
within a
study).A SpottedOwl researcher

not involvedin the
CaliforniaSpottedOwl meta-analysis
wastaskedwith
randomly selectinginformation from the databases
supplied by the respectivestudy-area researchers.
Tenrecordswererandomlydrawn from the capturehistory databasefrom each study area and 10 from
the reproductive database;individual researchers
were then requiredto provide papercopiesof the associatedoriginaldataformsor fieldnotes.At leastone
maleandonefemalewere drawnfrom eachage-class
tocheckthe survivaldatabase.
Therandomlyselected
informationwas then comparedwith the actualfield
data recordedon originalfield notes.If errorswere
found, an additional10 were randomlyselectedfor
checking.If errorswerefoundafterthe secondcheck,

DATA ANALYSIS
Direct

inferences

from

our results

are limited

to

studypopulationsanalyzedand time periodsduring

whichdatawerecollected.Inferences
beyondthestudy
populations
(e.g.totheCaliforniaSpottedOwl throughout the SierraNevada)arenot possiblewith thosedata
because
thestudyareasrepresentonlya smallfraction
of the totalareaof the SierraNevadarange,andthose
the entire database was checked for errors.
study areaswere not selectedrandomlyfrom a samAfter the first day of the workshopand prior to plingframeencompassing
theSierraNevada.
any data analysis,all researchers
were required to
Changesin analyticalmethodology
from previous
sign a certificationletter statingthat their data were SpottedOwl studies.-There has been considerable
correct, had been checked and rechecked, and were
debate over the most appropriatemeasureof the
ready for final analysis.Failureto signthe certifica- finite rate of populationchange(X) in SpottedOwl
tion would have meant exclusion of their data from
populations. Historically, Spotted Owl researchers
thefinalanalysis.Further,by signingthecertification, have estimated• using a Leslie projectionmatrix
researchers
explicitlyagreedthat their datacouldnot (XPM),
which was basedon estimatesof age- or
be withdrawn from the analysisafter results were stage-specific
survivaland fecundity(Franklinet al.


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTEDOWL
1996a,Caswell 2000). That method was the best avail-

able at the time for estimatingrates of population
change.Nevertheless,
the debateon ratesof population changein SpottedOwls using3.pM
has centered
on the centralissueof whether3.pM
is biasedbecause
the populationsare not geographicallyclosed (e.g.
thereare unknownratesof juvenileemigrationfrom
the study areas).If bandedjuvenile owls leave the
study area, live, and remain undetected,an estimate

of juvenilesurvivalusingmark-recaptureestimators
will be negativelybiased.For example,estimatesof
juvenile survivalprobabilitieson three study areas
for the Northern SpottedOwl increased42-137%
when they were adjusted,usingradiotelemetrydata,
for emigration from those study areas (Franklin et
al. 1999).Conversely,reproductivelyactive owls are
more likely to be detected than nonreproductively
active owls, which could result in an overestimate

(i.e. positivebias) of reproductiveoutput. If biased
survival or reproductiveoutput estimatesare used

in the projectionmatrix,estimatesof 3.p•would be
biasedas well. Thus,an importantissueconcemsthe

13

would not support a projectionmatrix approachfor
some of those demographic studies. Thus, we de-

cidedto rely on 3.t, whichestimates3. directlyfrom
the capture-recapturedata, to estimate changesin
owl numberswithin studyareas(Pradel1996,Nichols
and Hines2002).Inferencesand assumptionsrelevant
to that techniqueare explainedmore fully below and
in Appendix3.
Estimatingadult survivaL--The meta-analysisof
adult apparent survival was based on adult female
and adult malecapturehistoriesfor the five studyareas,where captureswere either initial captures,recaptures,or resightingsof color-bandedindividuals.We
definedapparentsurvival(0) as the probabilitythat
an owl alive in a particularyear t survivedto the same
time next year (t + 1) and remainedon the study area
and,hence,wasavailablefor recapture.The reciprocal
of apparentsurvivalwasa functionof both deathand
emigration.We assumedthat permanentemigration
of adult SpottedOwlsfrom studyareaswasvery low,
on the basisof data on Northern SpottedOwls (e.g.
Franklin et al. 1996b,Forsman et al. 2002). Hence, we

consideredapparent survival for California Spotted
Owls
to be an approximateestimateof true survival,
1996). With the exceptionof the SAB (essentially,a
geographicallyclosedpopulationfor which therewas the reciprocalof whichwasdeathonly.
Fromcapturehistoriesof individualsfirstcaptured
a goodestimateof juvenilesurvival),we couldnot be
certainthat we did nothavea biasedestimateof juve- as juvenilesor subadults,we removedencountersat

the youngerages,leaving only capturesat the adult
nile survival because of the likelihood of undetected
juvenileemigrationfrom the studyareas.Previously, age. Estimates of apparent survival and recapture
studiesin the SierraNevada (with the exceptionof probability(p,probabilitythatan animalalivein year
t is captured,recaptured,or resighted)were obtained
the LAS, wherejuvenilesurvivalwasestimatedfrom
with the Cormack-Jolly-Seber
model (Lebretonet
recapturedata) used a projectionmatrix based on
al. 1992) using Program MARK. The global model
estimatesof territorial owl survival and fecundity,
was(0xw
e Px.t*•),
where0 wasapparent
and a "surrogate"estimateof juvenilesurvivalwhich considered
survival probability,p was recaptureprobability,g
was "borrowed"from the SAB.The useof a surrogate
was study area, t was time (year), and s was sex.We
estimateof juvenile survival probablyintroducedan
assessed
goodness-of-fit
of this modelwith program
unknownbiasintotheestimates
of kp•because
of po- RELEASE(Burnhamet al. 1987).Assumptions
undertentialgeographicvariationin survivalrates.
lying use of mark-recapturedata for SpottedOwls
Despite those potential problems,we decided, and use of goodness-of-fitto evaluatethoseassumpduring the early stagesof the workshop,to estimate tionswasdiscussedin greaterdetailby Franklinet al.
kp• becausethere was sufficientdisagreementfrom (1996a).In general,studiesonCaliforniaSpottedOwls
someparticipantsin the workshopconcemingcom- were very similar in designto thosefor the Northern

plete exclusionof kpMfrom the analysis.Thus, our Spotted Owl. We estimated overdispersionin the
initial approachwas to estimateratesof population data using • = z2/df using the combinedchi-square
changeusingboth kp• (seeAppendix2) and a re- (Z2) valuesand degreesof freedom(df) from TEST2
centlydevelopedanalyticaltechniquefor estimation and TEST 3 from programRELEASE(Lebretonet al.

correctinference
to be takenfrom3.p•(Raphaelet al.

of 3.(referredto hereaskt;seePradel1996).Thatnew

1992). Estimates of • were used to correct estimated

methodwasemployedin aNorthernSpottedOwl metaanalysis(Franklinet al. 1999).We spentconsiderable
timeattemptingto estimatejuvenilesurvivalfromthe
capture-recapture data. However, we encountered
problemsin estimabilityof parametersfor juvenile
survival.In attemptingto solvethoseproblems,other
issuesconcerningbiasin estimatesof juvenilesurvival becameapparent(seealsoAppendix3). Eventually,

standard

errors

and Akaike's

Information

Criterion

(AICc)values(seebelow).Twenty-seven

modelswere
initially fit to the data from the five study areaswith

threestructures
(Ox.t•s,
Ox.•+s,
andOs.t)
onapparent
survival, and nine structures(pg+F pg+t+s, p[g+t]*s, pg*F p[g*t]+s'

Px*t•,
P•'P,s,andPrOonrecapture,
wherer wasannual

reproductiveoutput estimatedfrom the five study
areas. Program MARK (White and Burnham 1999)
we collectively
decidedthat estimates
of kp• would generatedthe log-likelihoodfunctionvalue, degrees
have someunknownbiasbecauseof thoseproblems, of freedom, and the small-samplebias-corrected
and analystsand most researchers
agreedthe data quasi-likelihood
AIC (QAIC•)(Sakamoto
et al. 1986,


14

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

Burnham and Anderson 1998) for each model evaluated.That criterionwas computedasfollows:

separatelybecauseestimateswere not comparable
acrossstudy areas(exceptfor the SIE and SKC study
areas) becauseof differencesin the field protocols
-2log Likelihood
2K(K+ 1)
used to estimate reproductive output (number of
QAIC
c=
•
+2K+ n- K--•fledgedyoungper pair). Raw data usedin the analysisfor eachstudy areaconsistedof numberof young
where K was the number of parametersestimated,
fledgedon a particularsite,site (territory)where the
• was the estimateof overdispersion,and n was the
young were detected,year, and age of the female
effectivesample size (i.e. number of binomial trials
(first-yearsubadult,second-yearsubadult,or adult),
includedin the likelihoodfollowingBurnhamet al.
for eachfemalemonitored.Prior to analysis,we di1987).ThesmallertheQAIC•valuefor a givenmodel,
vided the estimatesof reproductiveoutput for each
the better an approximationthe model was for the
site within eachyear by 2 to estimatefecundity,asinformationin the observeddata, given the set of
suminga 1:1sexratio.Weusedmixedmodelsbecause
models examined.
(1) individuals and territorieswere confoundedover

Usingthe minimumQAICcmodelfor p from the
initial 27 models,we fit the following 10 additional

time because the same females often bred on the same

models
forO:0d •)•.••)•+T,
•)g*•,
•)g+rr,
O•*Tr,
O••)T,•)rr'and

underestimate

•).,where T denoted a linear time trend, TT denoted a
quadratic time trend, and "." denoted a means only

territoryfor >1year;thatlackof independence
would
standard

errors

if methods

assum-

ing independencewere used (Franklinet al. 1999);
(2) modelingcould be conductedin a maximum-

model.Usingthe minimumQAICcmodelfromthose likelihood framework; (3) inferencewas made to sites
10 modelsthat includeda study area effect,we fit

ratherthan to separateoutcomes-year
by adjustments

additional
models:
(•latitudd
OSAB
......t' •s•c......• and•SAB,of the standarderrors;(4) the error covariancematrix

SKC
......• wherethe termrestdenoteda singlesurvival couldbe structuredappropriately;and (5) modelsalparameterestimatedfor the remainingstudy areas
lowedfor unbalanceddesigns(e.g.missingdata).
combined. Thus, we considered a total of 41 models
Raw datausedin theanalysiswereintegerdata(0,
(Table2).

1, 2, and rarely 3 or 4). Analysisof Northern Spotted
Owl fecunditydata showedthat variationin number
were computedas
fledgedwithin a year was proportionalto the mean,
which suggestsa Poissondistribution (Evans et al.
exp(-«
A/)
1993),althoughdata were not distributedas Poisson
zøi- R
(Franklinet al. 1999,2000). When we analyzedthe

California Spotted Owl data using mixed-model
r=l
analysisof variance,we relied on samplesizes that
were sufficientlylarge to justifynormal distributional
whereA• was the differencein QAIC• value for the assumptions.
On the basisof simulations,analysisof
minimumQAICcmodeland modeli, and R wasthe variance models were robust to severe departures
number of models (41 in our case)in the set. Model from normality (White and Bennetts1996). In addilikelihoodwascomputedastheQAICcweightfor the tion, analysisof variancemodelswere more robustto
modelof interestdividedby the QAICcweightof the data from discretedistributions,suchas the negative
bestmodel.Temporaland spatialprocessvariationof binomial,than was Poissonregression,
evenwhen it
apparent survival were estimated with the variance was correctedfor overdispersion
(White and Bennetts
components
moduleof ProgramMARK (White et al. 1996).Therefore,we decidedto rely on therobustness
2002, Burnham and White 2002). We distinguished of analysisof varianceto nonnormallydistributed
between
process
variation
(O2p
......)--thevariation
in a data, rather than relying on Poissonregression,to
given parameter,such as •)--and samplingvariation analyze the fecundity data, which have properties

TheAkaikeweights(BurnhamandAnderson1998)

Zexp(-«Ar)

similar
(var(•[0))--the

variation
attributable
to estimating

a parameterfrom sampledata (White 2000). Process
variation in populationparameterscan be decomposed into temporal processvariation (variation in
a parameter over time) and spatial processvariation
(variationin a parameteramongdifferentlocations),
whichrequiresthatsamplingvariationbe "removed"
from the total variation in the annual or spatialestimates of interest.

to a Poisson but are not distributed

as Poisson.

The mixed-modelproceduresalso allowed us to accountfor that dependenceof samplingvariationon
the mean(seebelow).As with Northem SpottedOwl
analyses(Franklin et al. 1999), we did not separate
individualbird effectsfrom territoryeffectsbecause
of the longevity of most individual females on
territories.

We used PROC MIXED (SAS Institute 1997) to fit

Estimatingfecundity.--We analyzed fecundity variousmodelsto estimatefecundityfor eachstudy
data for eachstudy areawith mixed analysisof vari- area. Initially, we followed the protocol developed
ance models (Rao 1997). We analyzed study areas duringtheworkshop.However,severalcomplications


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTEDOWL

15

TABLE
2. Descriptions
of mark-recapturemodelsexaminedin the meta-analysis
of adult apparentsurvival(t•)
of maleandfemaleCaliforniaSpottedOwlsfromfive studyareasin California.

Model

Description
of 0 structure

Description
of p structure

No effects

Studyareaand yeareffectswith interactions
Studyareaandyeareffectswith interactions
Studyareaandyeareffectswith interactions
Studyareaandyeareffectswith interactions

Studyareaeffect
Studyareaeffectwith additiveyeareffects
Studyareaeffectwith additivelineartime
effect

Studyareaeffectwith additivequadratic

Studyareaandyeareffectswith interactions

time effect

Studyareaandyeareffectswith interactions Studyarea,year,and gendereffectswith all
interactions

Study area and year effectswith interactions

Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions Additive studyarea,year,andgendereffects
Studyareaandyeareffectswith interactions Studyareaandyeareffectswith interactions
Study area and year effectswith interactions Annual reproductiverate effect
Study area and year effectswith interactions Annual reproductiverate and gendereffects
with interactions

Study area and year effectswith interactions
Studyarea and year effectswith interactions
Study area and year effectswith interactions

Studyareaandlineartimeeffects

Additive studyareaand yeareffects
Additive annualreproductiverate and
gendereffects
Additive studyareaand yeareffects
interactingwith gender
Studyareaandyeareffectswith interactions

with interactions

Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyarea,year,and gendereffectswith
all interactions

Studyarea,year,andgendereffectswith
all interactions

Studyarea,year,andgendereffectswith
all interactions

Studyarea,year,andgendereffectswith

Annual reproductiveoutput
Annual reproductiveoutputwith an additive
gendereffect
Studyarea,year,andgendereffectswith all
interactions

Study areaand year effectswith interactions
and an additivegendereffect
Additive study area and year effects
interactingwith gender
Additive studyareaandyeareffects

Additive studyarea,year,andgendereffects
Studyareaandyeareffectswith interactions
Annual reproductiverate and gendereffects
with interactions

Additive annualreproductiverateand
gendereffects
Annual reproductiverate andgendereffects
with interactions

Additive studyareaand yeareffects
interactingwith gender
Additive studyareaandyeareffects

all interactions

Studyarea,year,and gendereffectswith

Annual reproductiverateeffects

all interactions

Studyarea,year,andgendereffectswith
all interactions

Study area and year effectswith interactions


ORNITHOLOGICAL

16

MONOGRAPHS

NO. 54

TABLE 2. Continued,

Model

Descriptionof $ structure
Studyarea,year,and gendereffectswith

Descriptionof p structure
Studyarea,year,and gendereffectswith all

all interactions

interactions

Additivestudyarea,year,andgendereffects

Studyarea,year,and gendereffectswith
all interactions

Studyarea,year,andgendereffectswith

Studyareaandyeareffectswith interactions
and an additivegendereffect
Studyareaandyeareffectswith interactions

all interactions

Studyareaand quadratictime effectwith
interactions
Latitude

effect

Groupeffectof SABstudyareaversusother
studyareas
Groupeffectof SABstudyareaversusSKC
studyareaversusotherstudyareas
Groupeffectof SKCstudyareaversusother
studyareas
Year effect

Linear time effect

Quadratic time effect

Studyareaand yeareffectswith interactions
Studyareaandyeareffectswith interactions
Studyareaandyear effectswith interactions
Studyareaand yeareffectswith interactions
Studyareaandyeareffectswith interactions
Studyareaandyeareffectswith interactions
Studyareaand yeareffectswith interactions

arose,including(1) mis-specification
of the SAScode apparentthat someof the covariancestructureshad
used to run the models and (2) nonconvergence
of beenincorrectlycoded.In addition,somemodelsand
some of the models because there were too few incovariancestructuresanalyzedduringthe workshop
dividualsin the subadultage classeson most study failedto converge:
thelog-linearcovariance
structure
areas.In both the initial analysisand the subsequent for the LAS, SIE, and SKC study areas;and fixedre-analysis,we used a two-stage approach to fit effectmodels with age and year interactionsfor the
modelsto the data for eachstudy area (seeWolfinger ELD studyarea.That failure to convergewasbecause
1993). First, we used restrictedmaximum likelihood there were <3 subadults(first- and second-yearsubestimationwith model {age+ T; fecundity= 00+ 0• adult age-classescombined)for six, four, four, and
(age)+ •2 (year)lfor eachstudyarea,whereageof the eightyearson the ELD, LAS, SIE, and SKCstudyarbird was a categoricalvariable(either first-yearsub- eas,respectively.Therefore,we re-analyzedthe data
adult, second-year
subadult,or adult) and year was with correctlyspecifiedcovariance
structuresand we
a continuous variable. That model was executed with
usedonly adult femalesfor all the time-trendmodels

eachof four candidatevariancestructures:log-linear on all five study areas.Using data from only >3 year
variance (LOCAL = EXP(AGE YEAR), compound old females, the SAS code for each of the covariance
symmetric (CS), first-order autoregressive(AR1), structuresexaminedusingrestrictedmaximumlikeliand heterogeneous
first-orderautoregressive
(ARH1) hood estimation was:
PROC MIXED METHOD = REML;
(SASInstitute1997).We selectedthe mostappropriate

covariance
structureusingAIC, with only the covarianceparametersas the numberof parametersusedin

calculatingAICcbecauserestrictedmaximumlikelihood estimationignoresthe fixed effects(Wolfinger
1993).We usedthisstepto selectthe mostappropriate
covariancestructure for inclusion in the following
fixed-effectsmodels with fecundity as the response
variable:quadratictime trend (TT), linear time trend
(T), even-odd years (EO), linear time trend with an
additiveeven-oddyear effect(T + EO), and no time
trend (intercept,a meansor intercepts-only
model).
We used full maximum-likelihoodestimation(rather
than restricted maximum-likelihood estimation) to
analyzethosemodels;the numberof parametersin
that casewere the number of covarianceparameters
plusthenumberof fixedeffectparameters.
Themodel
that bestexplainedthe data for eachstudy area was

RANDOM

SITE YR;

REPEATED/ LOCAL = EXP(YR) SUB = SITE;
for log-linearvariance;
PROC MIXED METHOD = REML;
RANDOM YR;
REPEATED YR / TYPE = CS SUB = SITE;

for the compoundsymmetric;
PROC MIXED METHOD = REML;
RANDOM SITE YR;

REPEATEDYR / TYPE = AR(1) SUB= SITE;

for thefirst-orderautoregressive;
and
PROC MIXED METHOD
RANDOM SITE YR;

= REML;

REPEATEDYR / TYPE = ARH(1) SUB= SITE;
for the heterogeneous
first-orderautoregressive.
Again, we did not computea meta-analysisacross
the five studyareasbecauseof the differencein protoselectedusingAICc.After the workshop,it became colsusedto estimatereproductiveoutputin thefield.


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL

However,we madecomparisons
betweenthe SIE and
SKC studyareasbecausethosetwo areasusedsimilar
field protocols.In addition, we estimatedtemporal
processvariationin fecundityusingan intercept-only
modelfor eachstudyarea.
Estimatingrates of populationchange.--Weestimated the rate of populationincrease(•.) usingthe

17

approximatetrue survival(e.g.previousSpottedOwl
analysesin Burnhamet al. 1996,Franklinet al. 1999).
Thoseefforts reduce the movementasymmetryand
its effecton •-PM.However,thoseadjustmentsrequire
additional informationon emigration,suchas information from radiomarked birds. Second, the com-

puted •-PMis an asymptoticvalue expectedto result

temporal
symmetry
capture-recap•ture
model
ofPradel from the completeabsenceof temporalvariationin
(1996),whichwas denotedas •'RlSwhereRJSwas the vital rates,whereasthereis likely to be evidence
"eparameterizedJolly-Seber."
For notationalease,we

oftemporal
variation
inthedata.

Thus,
•M isacon-

stantvalue over a specifiedtime period whereas•-t
providesannual estimatesthat capturethe temporal
modifiedLeslieprojection
matrix!Franklinet al. variability in rates of populationchange.Third, val1996a) which can be denoted as •-PM where PM uesof fecunditymaybepositivelybiasedif nonbreedbirds are
denotes "projectionmatrix" and refers to a stage- ing birds are not detectedor if unsuccessful
birds.The fourth
structuredprojectionmatrixapproach(Caswel12000). not detectedasreadilyas successful
reason is related to the first and involves the fact that
Thetwo typesof •. differin theirdefinitionsand interestimates
of juvenile survival are probably negapretations,aswell asin theirmethodsof computation.
Here, we presenta brief discussionof thosedistinc- tively biasedwhen they are obtainedusingcapturerecapturemethods (Franklin et al. 1999). That is of
tions and differences.
Variable
•t estimates
•'etherateofchange
inpopu- concernwith California SpottedOwls becauseof
the paucityof data for estimatingjuvenile survival.
lationsizebetweenyearst and t + 1:
In summary,the •-t should provide reasonableestimatesof annualratesof changein abundanceof ter•'t=Nt+l
ritorialbirds on the studyarea.The )•PMis perhaps
It
best viewed as an abstractionto the extent that (1) it
is an asymptoticquantity that assumesno temporal
whereNt is abundanceat year t. In the caseof the variation in vital rates, and (2) it includes all losses
California
Spotted
Owlanalysis,

abundances
and•t from, yet not all gainsto, the population(no recruitapply to subadult and adult territorial owls on the ment from outsidethe study area is includedin that
study areas.That rate of changein abundanceis a quantity).
functionof the four fundamentaldemographicvariThereare severalassumptions
underlyingestimaables:survival rate, reproductiverate, emigration, tion of •'t that needto be considered
(seeHinesand
and immigration.Thu^s,demographicexplanations Nichols 2002 and Franklin 2002 for more complete
for specificvaluesof •-t requireadditionalinforma- details).First,interpretationof suchestimatesis most
tion on those•fundamental
demographic
variables.
straightforwardwhen studyarea sizeand boundary
Variable kPM is computedfrom projectionma- configurationsremain unchangedthrough time. If
trices parameterizedwith means of time-specific studyareasexpandor contractovertime,the resultestimates,or constant-parametermodel estimates, ing •-t will reflectthat the populationto which inferfor stage-specific
survivalandfecu?dityfor juvenile, encesarebeingmadeisalsoexpandingor contracting.
subadult,and adult survival.The •'PMresultingfrom Second,all animalswithin the study area must have
thosecomputations
represents
theasymptoticgrowth someprobabilityof beingrecapturedthroughoutthe
ratefor a populationexposedto theprojectionmatrix study.If portionsof the study area are inaccessible
vital ratesyearafteryear.Thatvaluecanbe viewedas during some years of the study, but then become
a functionof the averagevital ratesbut is not neces- accessible
for trappingin subsequent
years,individusarily a goodestimateof the averagerate of change als capturedin the inaccessible
portionof the study
in numberof birds on the study areafor at leastfour area will suddenly become "new recruits" to the
reasons.First,thereis an asymmetryin the way move- populationeven though they had been present,but
ment is treated in vital rates representinggains and not availablefor sampling,in previousyears.Third,
losses.New individualsare addedto the projected permanenttrap responsein captureprobability can
populationonly via in situreproduction,as reflected biasestimatesof •'t (Hinesand Nichols2002).When

in the fecundity estimates.However, SpottedOwl animalsrespondpositivelyor negativelytobeingcapsurvivalestimatesrepresentapparentsurvivalin that tured (Seber1982),a differencein captureprobability
their complementsincludeboth death and perma- occursbetween animals that have, and have not, been
nent emigrationfrom study areas.Thus, lossesfrom captured previously and marked. Permanent trap
the populationoccurvia both deathand permanent responsein the standardCormack-Jolly-Seber
modemigration. Note that sometimesefforts are made to els induces no bias in survival estimates (Pollock et
adjustapparentsurvivalestimatesso that they better al. 1990),but estimatesof populationsizeunder the

denoted
ayear-specific
RJS
estimator
as•t foryeart.

Prior analysesof SpottedOwl data have used a


18

ORNITHOLOGICAL

MONOGRAPHS

NO. 54

Cormack-Jolly-Seber
model are biasedin the face of
permanent trap responsebecausethe differencein
capture probability betweenmarked and unmarked
animals causespredictableproblems(Nichols and
Hines 1984).That samebiasalsoappliesto estimatesof

often include the specificationthat all emigrationis
permanent.As notedby Kendallet al. (1997),random
temporaryemigrationled to unbiasedestimates
of the
sizeof the "superpopulation,"
consistingof birdshaving somechanceof beingin the sampledareaduring
3,•.HinesandNichols(2002)foundthatbiasis positive the samplingperiod.In thecaseof randomtemporary
in thepresence
of a trap-happyresponse
and negative emigration,
we expected
estimates
of3,t tobeunbiased
in the presenceof trap-shyresponse.That biasis not for changesin superpopulation
size.However,nonsubstantialfor smalllevelsof trap responses
but could random(e.g.MarkovJan)formsof temporaryemigra-

of 3,t,andwe were
beif levelsof trapresponse
arehigh.If trapr•esponsetion couldleadto biasedestimates
of
changesovertime,thenmisleadingtrendsin •t could not aware of investigationson the consequences
result.Fourth,estimates
of3,t arebiasedin thepresence suchtemporaryemigrationto estimation.
of heterogeneouscapture probabilitiesamong indiTheprimaryinferencefrom )•M regardingpopulaviduals or unidentified
classes of individuals.
Hines
tion rate of changewas to the territorial owls on the
and Nichols (2002)show that heterogeneous

capture s•tudyarea. Previously,the primary inferencefrom
probabilitiesdo not biasestimatesof 3,twhen popula- )•M had been phrased as "did the territorial owls
tion growth rate is modeledas a singleconstantover on the study area replace themselves?"(Burnham
all time periods.Smallbiasdoesoccurwhen estimat- et al. 1996,Franklin et al. 1999). However, that inferingtime-specific
3,c However,thatbiaswasnotassub- ence provided no information on where replacement
stantiala problemas that resultingfrom permanent owls might go with respectto study area boundartrap response.
ies. In other words, the inferenceapplied if all of the
Violation of the first two assumptionsdo not pro- young producedon a study area remained on that
ducebias,in thattheestimatorof 3,t is not performing area and then exhibited similar survival and fecunas it was intended (Hines and Nichols 2002). When the

dity rates of adults on that area. Thus, we rephrased

study areachanges,the estimatedpopulationchange
is the result of two conceptuallydistinct processes.
The first processinvolves expansion of the study
area and increasein number of animalsexposedto
samplingthat resultfrom that expansion.The second
process
involveschangesin thenumberof animalson
the sampledarea;that is the changeof interestand the

that inference

oneto whichwe wouldlikeestimates
of 3,tto apply.

as "would

the territorial

owls on the

studyareareplacethemselves
if the systemwasge9graphicallyclosed?"In contrast,the inferencefrom
included recognition that the system may not be
geographicallyclosedand was phrasedas "were the
territorial owls on the study area being replaced?"
(Franklin et al. 1999). That inferencewas about the
owl populationsresidingwithin a specificstudyarea.

Other assumptions underlying open capture- An advantage
of 3,t wasthat estimates
of juvenilesurbothimmigrationand
recapturemodelshavenot been^specifically
investi- vivalwerenotrequiredbecause
gatedwith respectto effectson )•t' For example,we emigrationwere accountedfor by changesin number
assumedno tag loss and no tag-induced mortality.
Becausewe had no reasonto suspectthat thosewere
important problems, we did not investigate consequencesof their violation.However,lossof the same

type of bandsusedon 3,788NorthernSpottedOwls
was only 0.1% (Forsman et al. 1996). In addition,
we recapturedowls when color band combinations
becamedifficult to read. Similarly, homogeneityof
demographicrate parameters(e.g. survival) among
individuals is assumed in open population capture-recapture models.Our focuson territorial birds
eliminated the potentially large variation between
territorial and "floater" birds, and we did not know

the consequences

of remainingvariation in param-

of owlsovertime.Thus,thepotentialbiasfromimpreciseand inaccurateestimatesof juvenile survival was

avoided.Ho•wever,a primarylimitationof inferences
concerning)•t was that it was not possibleto estimate
the relativecontributionsof the differentcomponents
to populationgrowth(e.g.reproduction,
immigration,
death, and emigration)without additional data. For
example,immigration could sustaina demographically (basedon survival and reproduction)declining
population(i.e. the populationcouldbe a sink;sensu
Pulliam 1988).It shouldbe noted that most Spotted
Owl populations,as defined by the usual scale of
study,werelikely maintainedby immigration(behavior commonlyattributedto sink populations),while
alsosupplyingrecruitsto otherpopulations(behavior
commonlyattributed to sourcepopulations).Thus,
the source-sinkdichotomymay not be as usefulwith
SpottedOwl popula•tions
as with someotheranimal
populations.Thus,•t WaSan importanttool but will
not sufficeas a singleassessment
of the health of a

eters among individual territorial birds. As with
most explorationsof heterogeneous
rate parameters,
we suspectedthat substantialvariation could lead to
important bias, whereas relatively minor variation
would be lessof a problem.The high annualsurvival

estimatesfor SpottedOwls did not permitsubstantial
heterogeneity(i.e. it would not be possibleto have SpottedOwl population.Consequently,all relevantinsuchhighmeansurvivalif manyindividualsexhibited formation should be used to draw inferences about the
greatly reducedsurvival). Open model assumptions stability of California SpottedOwl populations(see


POPULATION DYNAMICS OF THE CALIFORNIA SPOTTED OWL

19

below).For example,estimatesand trendsin survival populationsizewascomputedasN• = nJp•(wherenx
and fecundityratesshouldalsobe evaluatedto assess is the number of birds caughtin initial year x) and
thehealthof a SpottedOwl population.
thefirstrecruitment
rateas

To estimatek• for eachstudyarea,we employe•d assumingPx=•x+•- That latter assumptionwas
random effectsmodels in MARK that used the

neededbecausewe requiredan initial abundancefor

simulations
but wereunableto estimate
from model{•e p•,)•} as the basisfor the analysis. demographic
Model {•t, Pt,k•},whichwasusedasthe basisfor the the initial captureprobability.That particularsolution
random-effectsmodels,allowed •, )•, and p to vary (equatingthe first and secondcaptureprobabilities)
by year;noneof the parameterswere otherwisecon- seemedreasonableto us, although other solutions
strained.We considered
the followingrandom-effects to that sameproblemhave been usedby others(e.g.
models:lineartrendin )• (T), quadratictrendin )• (TT), N1 = N2, Jolly 1965;p• = 1, Schwarzand Arnason
and mean )• acrosstime (.). Those three models were

1996).Foreachsimulation,
we usedmodel•ep•,)• to

considered

computeestimatesof )•tand, from those,At. We ran

with and without

elimination

of the first

estimable
k•because
ofpotentialbiasesdueto trapre-

1,000 simulations; from that distribution of simula-

sponse,perhapsexacerbatedby a "learningcurve"on tions,we computed95%confidenceintervalsfrom the
the part of observers.
Because
of the possibilityof dif- ithand
jthvalues
of At, where
i =(0.05)(1000)
andj =
ferentcaptureprobabilities
for markedandunmarked (0.95)(1000).

We alsoconducteda meta-analysis
usingonly the
birds,we tendedto disregardthefirstestimable
)•tand
to focuson modelsthat did not include that param- studyareasin the SierraNevada(LAS,ELD, SIE,and
eter.Prior to conductingthe analysison )•t for each SKC) to examine potential correlationsin annual
studyarea,we adjustedsomeof the studyareadeft- variation among thosestudy areas.We considered
nitions to meet the assumptionof a geographically fixed-effects
models
)¾ )•x.eandk•+•,whereg was
consistentarea where birds had some probability study area and with • and p structuredas g * t in all

of being surveyed.For the ELD, only the capture- threemodels.
Wealsoconsidered
)•t•without
thefirst
andsecondestimable
)•t;withoutthefirst,second,and
last)•t;withoutthe firstand lastk•;and withoutthe
first )•t'In the analysisof the individualstudyareas

recapturedata from the smaller density study area
wereused;for the SIE and LAS studyareas,a subset
of the capture-recapturedata from a smaller geographicarea were used. Estimatesof overdispersion
(•) were recomputedfor the capture--recapture
data
usingprogramRELEASEand usinga globalmodelof

{%• P,*•)•.t}for eachstudyarea,with an interaction
betweensexand year for •, p, and )•.

and the meta-analyses
acrossthe four studyareas,we
correctedfor overdispersion,and we used the same
model selectionapproachthat was used to estimate
adult survival.

Comparison
of Sierraand Sequoia
and KingsCanyon

parksstudyareas.--The
SIEandSKCrepresentTo makethe annualestimates
of )5 moreinterpre- national
table, we translated those estimates into estimates of

realizedchangeof thepopulations
on eachstudyarea.
Annualrealizedchanges
wereestimatedastheproportionof theinitialpopulation(i.e.in theinitialyearused
for analysis)
remainingin yeart (i.e.At = N,/N•wherex
is theinitialyear).Therefore,realizedchangeprovided
theestimatedtrajectoryof the populationoverthetime
periodfor which)• wasestimated,withoutrequiring
estimationof numbersof owls on eachof the study
areas.Realizedchange(At)wasestimated
as:

ed paired study areaswhere SIE had been managed

for timber productionand SKC had been managed
as a nationalpark. They coveredsimilar rangesin
elevationand were in closeproximity(19 km), minimizing differencesin weatherpatterns.Also,investigatorsusedthe samefield protocolsfor reproductive
output on the two study areas,which allowed direct

comparison
of fecundityrates.We usedeffectsizesto
comparethe differences
betweenthe two studyareas

in termsof thethreedemographic
parameters:
adult
fecundity,adult survival,and mean )•.
RESULTS

i=x

ADULT SURVIVAL

wherex wasthe firstestimated)•t-For example,if
was 0.9, 1.2, and 0.7 for years 1993, 1994, and 1995,

We used 975 marked adults in the analysis

respectively,
thenAt fort = 1996
was(0.9)(1.2)(0.7)
= of apparent survival (171 from the ELD, 223
0.756indicatingthat 75.6%of the startingpopulation

remainedafterthreeyears.

from the LAS, 307 from the SAB, 168 from the

Tocompute
95%confidence
intervals
for At,we SIE, and 106 from the SKC). From program

useda bootstrapalgorithm.Specifically,
we computed RELEASE,the overall goodness-of-fitfor the
time-specific
recruitmentrates•) for eachstudyarea global model was 3(2= 184.74with df = 167
usingmodel•t, Pt,ft in ProgramMARK. Usingthose (P = 0.165),indicatingthat the globalmodel fit
estimates, we simulated data inMARK where the initial

the capture-recapture
data. The resultsof that