Ornithological Monographs 40

(ISBN: 0-943610-50-8)

PATTERNS

AND

SIGNIFICANCE

OF GEOGRAPHIC

VARIATION
GROUP

EVOLUTIONARY

IN

THE

OF THE

SCHISTACEA

FOX

SPARROW

(PASSERELLA ILIA CA)

BY

ROBERT

M.

ZINK

Museum of Vertebrate Zoology and
Department of Zoology
University of California
Berkeley, California 94720

ORNITHOLOGICAL

MONOGRAPHS
PUBLISHED

THE

AMERICAN

BY

ORNITHOLOGISTS'

WASHINGTON,
1986

NO.

D.C.

UNION

40


PATTERNS

AND

SIGNIFICANCE
VARIATION
GROUP

EVOLUTIONARY

OF GEOGRAPHIC
IN THE

OF THE

FOX

SCHISTACEA
SPARROW

(PASSERELLA ILIA CA)


ORNITHOLOGICAL

MONOGRAPHS

This series,publishedby the American Ornithologists'Union, hasbeen establishedfor major paperstoo longfor inclusionin the Union'sjournal, The Auk.
Publicationhasbeen made possiblethroughthe generosityof the late Mrs. Carll
Tucker and the Marcia Brady Tucker Foundation, Inc.

Correspondenceconcerningmanuscriptsfor publication in the seriesshouldbe
addressed
to the Editor, Dr. David W. Johnston,Departmentof Biology,George
Mason University, Fairfax, VA 22030.

Copiesof OrnithologicalMonographsmay be ordered from the Assistantto
the Treasurerof the AOU, Frank R. Moore, Departmentof Biology,University
of SouthernMississippi,SouthernStation Box 5018, Hattiesburg,Mississippi
39406. (Seeprice list on back and inside back covers.)
OrnithologicalMonographs,No. 40, viii + 119 pp.
Editor of Ornithological Monographs, David W. Johnston

SpecialReviewersfor this issue:Dennis M. Power, SantaBarbaraMuseum
of Natural History, 2559 Puesta Del Sol Rd., Santa Barbara, CA;
RichardF. Johnston,Museumof Natural History, Universityof Kansas, Lawrence, KS

Author, Robert M. Zink, Museum of VertebrateZoologyand Department
of Zoology,University of California, Berkeley,California 94720. Present address:Museum of Zoology, Louisiana State University, Baton
Rouge, Louisiana 70803
First received,10 September1984;accepted12 August 1985; final revision
completed 4 May 1986
Issued December 9, 1986

Price $15.00 prepaid ($12.50 to AOU members).
Library of CongressCatalogue Card Number 86-72776
Printed by the Allen Press,Inc., Lawrence, Kansas 66044
Copyright ¸ by the American Ornithologists'Union, 1986
ISBN: 0-943610-50-8


PATTERNS

AND

SIGNIFICANCE

VARIATION
GROUP

EVOLUTIONARY
OF GEOGRAPHIC

IN THE
OF THE

SCHISTACEA

FOX

SPARROW

(PASSERELLA ILIACA)

BY

ROBERT

M. ZINK

Museum of Vertebrate Zoology and
Department of Zoology
University of California
Berkeley, California 94720

ORNITHOLOGICAL

MONOGRAPHS
PUBLISHED

THE

AMERICAN

BY

ORNITHOLOGISTS'

WASHINGTON,
1986

NO.

D.C.

UNION

40



TABLE
LIST

OF FIGURES

LIST

OF TABLES

LIST

OF APPENDICES

OF CONTENTS

...............................................................................................................................................
vi

...................................................................................................................................................
vii

INTRODUCTION

.......................................................................................................................................
viii

..........................................................................................................................................................
1

GEOGRAPHICVARIATION; SIGNIFICANCEOF PATTERNS,EVOLUTIONARY
INFERENCES,
AND GOALSOF ANALYSIS.................................................................................
1
OBJECTIVES OF THE PRESENT STUDY .....................................................................................................
3
PREVIOUS STUDIES OF GEOGRAPHIC VARIATION IN FOX SPARROWS ............
5

STUDY SITES, SAMPLING DESIGN AND TECHNIQUES, AND BRIEF
SUMMARY
MATERIALS

OF NATURAL
AND

METHODS

HISTORY

...................................................................................
6

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

10

ELECTROPHORESIS .................................................................................................................................................
10
MORPHOLOGY

........................................................................................................................................................
12

Study skin measurements ..................................................................................................
12
Skeletal

measurements

.............................................................................................................................
13

Numerical analysis of skin and skeletal characters ...........................................
13
R•NDOMNESS

IN GEOGRAPHIC PATTERNS; MANTEL

TESTS ......................................
15

RESULTS
...................................................................................................................................................................
16

ELECTROPHORETIC ANALYSIS ..........................................................................................................................
16
Locus level ................................................................................................................................................
16
Individual level .......................................................................................................................................
17

Population level .....................................................................................................................
24
Genetic distance between population samples ........................................................
25
Geographic pattern of protein variation ..........................................................................
25
F-statistics and the analysis of genetic structure among populations

....................................................................................................................................................
29

Relationships among estimates of genetic variation and their environmental and geographiccorrelates.................................................................
29
Analysis of levels of gene flow ..........................................................................................
32
Test of the neutrality hypothesisfor allelic polymorphisms within
populations .....................................................................................................................
33
MORPHOLOGICAL

VARIATION

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

33

Univariate character analyses ..............................................................................................
33
Multivariate analysis of variance ......................................................................................
47
Principal components analysis...........................................................................................
48
Cluster analyses .....................................................................................................................
58
Mantel

tests ..............................................................................................................................................
66

Environmental and morphological variation ............................................................
69
DISCUSSION

.....................................................................................................................................................................
70

POPULATION GENETICS OF Fox SPARROWS: EMPIRICAL RESULTS AND
PATTERNS ............................................................................................................................................................
70


Levels and the nature of protein variation within populations ........ 71
Environmental and phenotypic correlates of protein variation ........ 73

Relationshipsamong geneticcharacteristicsof populations .................
73
Genetic variation and its relationship to population demography
and subspeciesdistributions .........................................................................................
74
Levels and patterns of geneticvariation among populations ..............75
POPULATION

GENETICS OF FOX SPARROWS: INFERENCE OF PROCESSES .....

77

Recency of common ancestry ............................................................................................
77
Gene flow and effective population size ..........................................................................
78

Rates of molecular evolution ..........................................................................................................
79
Natural selection .........................................................................................................................................
81

Summary .................................................................................................................................
81
MORPHOLOGICAL VARIATION

.........................................................................................................................
82

Levels of character variation: Systematic and ecologicalconsid-

erations ..............................................................................................................................................
82

Levels and patterns of character variation among populations ........ 86
Potential environmental determinants of morphological diver-

gence ...................................................................................................................................
87

Historical patternsand temporal stability of phenetic relationships
among populations ......................................................................................................
89
MORPHOLOGICAL AND PROTEIN COVARIATION

........................................................................
91

A MOLECULAR PERSPECTIVE ON THE ORIGIN OF MORPHOLOGICAL VARIA-

TIOn4 ..................................................................................................................................................................
92
A HYPOTHESIS FOR THE ORIGIN AND MAINTENANCE OF MORPHOLOGICAL
DIFFERENCES AMONG FOX SPARROW POPULATIONS .............................................
93
EVOLUTIONARY

SIGNIFICANCE OF GEOGRAPHIC VARIATION

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

95

GENERAL CONSIDERATIONS

...................................................................................................................
95

GEOGRAPHIC

AND ADAPTATION:

vARIATION

AN INDIRECT

AS-

SESSMENT ........................................................................................................................................................
95

GEOGRAPHIC vARIATION AND SPECIATION ........................................................................
96
CONCI•USION .........................................................................................................................................
102
TAXONOMY AND THE SEARCH FOR EVOLUTIONARY TAXA ........................................
102
ACKNOWLEDGMENTS

SUMMARY

..............................................................................................................................................................
105

LITERATURE

APPENDIX

........................................................................................................................................
105

CITED

.............................................................................................................................................
107

I ............................................................................................................................................................
118

LIST

Figure 1.
2.

3.
4.
5.

6.

OF HGURES

Breeding rangesof the 18 subspeciesof the Fox Sparrow ....................
4
Location of the 31 collecting sites ...................................................................................
7
Breedingdistribution of seven subspeciesof the Fox Sparrow ...... 8
UPGMA phenogram of genetic distances...................................................................
28
Examples of gene flow levels ..........................................................................................
32
Analysis of gene flow in the Fox Sparrow ..................................................................
33


7.

Expected and observed distributions of alleles ....................................................
34

8.

Variation

in cube-root

of mass for males ....................................................................
35

9. Variation in wing length ...............................................................................................
38

10. Variation in bill width and tarsuslength .....................................................................
39
11.

Variation

12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.

Variation in width of the skull and length of the tibiotarsus ............40
Variation in lengths of coracoid and posterior synsacrum...................
41
Variation in width of synsacrum and length of sternum ........................
41
Means of bill width, length, and depth againstcube-root of mass 46
Means of five skeletal charactersagainst cube-root of mass ..............
47

Principal componentsanalysis of male skin characters............................
48
Principal componentsanalysisof female skin characters......................
50
SS-STP analysisof PC skin scoresfor males ..........................................................
51
SS-STP analysisof PC skin scoresfor females ....................................................
53
SS-STP analysisof PC II scoresfor skin charactersfor males ....... 54
Pie diagrams of PC scoresfor males (III) and females (II) ..................
55
Principal components analysisof male skeletal characters..................
56
Principal componentsanalysisof female skeletalcharacters............56
SS-STP analysisof PC skeletalscoresfor males ................................................
57
SS-STP analysis of PC skeletal scoresfor females ...........................................
58

in hind toe and outer rectrix

..........................................................................
40

27.

PC II scores for skeletal characters

28.
29.

30.
31.
32.

SS-STP analysisof PC III scoresfor male skeletalcharacters........ 60
UPGMA phenogramoftaxonomic distancesfor males, skin data 62
WPGMA phenogramfor female taxonomic distances,skin data 63
UPGMA phenogram for male skin characters.....................................................
64
UPGMA phenogram of taxonomic distancesfor males, skeletal

for males ........................................................
59

data ...............................................................................................................................................................
66

33.

UPGMA phenogram of taxonomic distancesfor females, skeletal

data ...............................................................................................................................................................
67

34.

UPGMA phenogram for male skeletal characters ...........................................
68

LIST

Table

1.
2.
3.
4.
5.
6.
7.
8.

OF TABLES

Description of study sites ..............................................................................................
9
Aspects of the breeding distribution of Fox Sparrows ................................
10
Electrophoretic conditions used for study of Fox Sparrows ................
11
Number of alleles per locus, observed and expected frequencies,
and numbers of heterozygotes.......................................................................................
17
Allelic frequencies for variable loci ..................................................................................
18
Genetic distancesof Nei and Rogers between samples..............................
26
Fs, analysisof electrophoreticdata ..............................................................................
29
Correlations between aspectsof genetic variation .............................................
30

9.
10.

ANOVA
for skin characters ...........................................................................................................
36
Correlations between skin characters .................................................................................
37

11.
12.

ANOVA
for skeletal characters ..................................................................................................
42
Correlations between skeletal characters ........................................................................
44

13.

Character correlations with principal components, skin characters 49
vii


14. ANOVA of individuals' scoreson principal components ......................
52
15. Character correlations with principal components, skeletal characters

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

55

16.
17.
18.
19.
20.

Character correlations with PC I for 10 separatePCAs ............................
61
t-values resulting from Mantel tests ...............................................................................
69
Canonical variable analysis of skin and environmental data .............
70
Canonical variable analysis of skeletal and environmental data ... 71
Summary of coefficientsof variation for several passedfiespecies 83

21.

Coefficients

of variation

LIST
I.

for skeletal characters

....................................................
84

OF APPENDICES

Locations of Sample Sites ..............................................................................................................
118

viii


"... one of the basic problems in evolutionary biology is to explain the nature and
origin of the differencesbetween different populations of the same species."
(Merrell 1981)

INTRODUCTION

GEOGRAPHICVARIATION: SIGNIFICANCEOF PATTERNS,
EVOLUTIONARYINFERENCES,
AND GOALSOF ANALYSIS

Merrell's remark nicely illustratesthat the studyof intraspecific,or geographic,
variation can contribute to the understandingof evolutionary processes(Mayr
1980). In fact, Gould and Johnston (1972: 457) stated that "the foundation of
most evolutionary theory restsupon inferencesdrawn from geographicvariation
or upon the verification of predictionsmade about it." The evolutionary significance of geographicvariation traditionally rests upon two assumptions.First,
natural selectionis thoughtto increasethe degreeto which populationsare adapted
to locally differing environments. Hence, a pattern of geographicvariation can
indicate a seriesof adaptive responses'togeographicallyvarying selectionregimes.
Secondly,many biologistsbelieve that the processsof geographicdifferentiation
is also a model of the origin of species.That is, speciation is usually envisioned
to consistof the conversionof geneticvariation from within to among populations

coupled with the origin of reproductive isolation (Mayr 1942, 1963, 1970). At
the least, analysisof geographicvariation might clarify the nature of phenotypic
and genotypicchange,and possiblythe evolution of reproductiveisolation (Zink
and Remsen, in press). These basic assumptionsabout the evolutionary significance of geographicvariation are not without challenge. Differential patterns of
geneflow, constrictionsin effectivepopulation size, and random geneticdrift can
generategeographicpatternsof variation in the absenceof natural selection(Rohlf
and Schnell 1971; Lande 1985). There is also some opposition to the classical
notion that speciation is merely an extension of the processof infraspecific differentiation (Goldschmidt 1940; Eldredge and Cracraft 1980; Cracraft 1983).
Nonetheless,whether or not one acceptseither assumptionor both of them, study
of geographicvariation is of value becauseit might exposeaspectsof the processes
of adaptation and speciation.
A primary objective in the analysisof geographicvariation is to identify patterns
of variation and explain their evolution. In recent years both the methods and
geographicscaleof analysishave changed.New methods involve types of data
gatheredand techniquesand theories of data analysis.Biochemical tools are being
used with increasingfrequency to study the geneticsof the microevolutionary
process(Barrowclough1983). Quantitative,computer-assisted
analyseshavegreatly
improved the description of patterns of geographicvariation. In particular multivariate statistical methods have been widely employed because, as eloquently
statedby Sokal and Rinkel (1963), "Geographicvariation is not likely to be due
to adaptation of a few charactersto a singleenvironmental variable, but is doubtlessa multidimensional processinvolving the adaptation of many charactersto
a variety of interdependentenvironmental factors whose gradients and ranges
probably overlap in a rather complex fashion." Implicit in the characterization
of geographicvariation by Sokal and Rinkel is a messagethat the types of traits
often surveyedfor geographicvariation might have complex genetic bases,re-


2

ORNITHOLOGICAL

MONOGRAPHS

NO. 40

quiring quantitative geneticanalysesto distinguishgeneticvs environmental contributions to phenotypicvariation as well as the geneticresponsesto potentially
antagonisticforcesof natural selection.Although descriptionand analysisof variation in traits traditionally consideredwill continue to be of value, quantitative
geneticstudies(Price et al. 1984b) of polygenictraits, comparisonof geographic
differencesin ontogenetic"trajectories" (Alberch et al. 1979), and analysisof
biochemical charactersfor which the genetic basis of variation is known (e.g.,
Barrowclough1983), will be necessarydirectionsfor future studiesof geographic
variation. In addition, even time-honored definitions of speciesand speciation
should be evaluated (eracraft 1983).
An important objective of studiesof geographicvariation is to determine the
extent of population subdivision or differentiation. In other words, on a continuum
from panmixia to complete subdivision and cessationof gene flow, what is the
genetical population structure of a particular species(no matter how speciesare
defined)?The nature of population structureinfluencesthe processesof adaptation
(Wright 1978)and speciation(Templeton 1980b;Slatkin 1985b).Hence,empirical
estimatesof populationstructureare of interest.Traditionally the extentof genetic
differencesamong populations was inferred from the extent of differentiation in
external morphology. Electrophoretic analysisof enzyme loci has provided a tool
for documenting geneticvariation in natural populations (Lewontin 1974; Ayala
1976, 1982; Smith et al. 1982), although relatively few surveys of avian species
exist (Barrowclough 1983; Barrowcloughet al. 1985). In contrast to traditional
types of charactersanalyzed in studiesof geographicvariation, one can determine
an individual's genotypeat each of up to 100+ loci (the limits on the number of
loci are as much financial and logisticalas they are technical).These geneticdata,
analyzed in light of quantitative predictions of population genetic models, can
elucidate the genetic structure of populations, levels and patterns of gene flow,

effectivepopulation sizes,strengthand nature of natural selection,and the pattern
of evolutionary divergenceof populations and species.A further advantage of
molecular charactersis the approximately uniform, time-dependent rate of evolution, a "molecularclock," which allowsa temporal perspectiveon the divergence
of groupsof individuals. Thus, becauseevolution ultimately consistsof genetic
change,biochemicalmethods that exposethe geographyof geneticvariation are
of considerableinterest if the goal is to estimate the pattern and timing of the
fragmentation of populations and patterns and mechanismsof speciation.
Analyses of covariation of morphology and proteins indicate whether they
evolve in concert or are "decoupled" (Schnell et al. 1978; Patton et al. 1979;
Smith 1981; Yoshiyama and Sassaman 1983), although Lewontin (1984, 1986)
suggestssome cautionsto be used in interpreting patterns of covariation. Significant genetic divergencehas been found where an absenceof morphological differentiation would have been interpreted previously as evidence of a lack of
population structure. Conversely, morphological patterns of variation might not
reflect the historical genealogyof populations becauseof nongenetic environmental influences (Chernoff 1982). In birds, morphological differences among
populations are often accompaniedby no or few detectabledifferencesat enzyme
loci, unlike some other vertebrates (Barrowclough 1983). Nevertheless, the goal
is to explain how and why populations have their particular sets of phenotypic
and genotypicattributes.Hence, all data setshave relevance,especiallywhen the


GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

3

strengthsand weaknessesof each are recognized.Biochemical and morphological
data setsshould be viewed as complementary.
Some recent studiesof geographicvariation have included assessmentof "microgeographic"patternsof variation (Patton and Feder 1978, 1981; Chesser1983).
This trend toward documentationof microgeographicpatternsstemsfrom a desire
to discover the smallest aggregationof individuals in nature that might function
as an evolutionary unit (Ehrlich and Raven 1969; Jackson and Pounds 1979;
Cracraft 1983). For example, if all individuals from a specieswere studied, one
could pool individuals, beginningwith a randomly drawn one, until either genetic
and/or morphological gapswere encounteredor all specimenswere in one group
(a taxon without infraspecific variation). A gap could be quantitative, such as
nonoverlappingrangesin some trait, or a medhstic,discretephenotypic state. If
consistentgroupingsof individuals obtain, whatever their taxonomic designation,
it is then necessaryto discovertheir evolutionary relationshipsand how and why
their pattern of relationships arose.
An extensiveornithologicalliterature existson patternsof geographicvariation
in a variety of traits (Zink and Remsen,in press).An examination of this literature

has allowed evaluation of some important topics in evolutionarybiology, such
as local adaptation, hybridization, sexualdimorphism, ecologicalisolation, Pleistocene speciation,and others (Johnson 1980). Study of patterns of geographic
variation in birds contributed to the development of the theory of allopatdhc
speciation (e.g., Huxley 1942; Mayr 1942; Streseman 1975). However, recent
treatises (Endler 1977; White 1978; Bush 1982) on speciation were constructed
largely without input from modern analysesof avian geographicvariation. Ornithologistswill continue to contribute to these researchareas, especiallywhen
newer methods of data gatheringand analysisdiscussedabove are used to complement traditional typesof studies.
OBJECTIVES OF THE PRESENT STUDY

The goal of this study is to determine what evolutionary forcesinfluence the
historical origin and current maintenance of geographicvariation in the Fox
Sparrow (Fig. 1). The analysiswas restrictedto some members of the Schistacea

group (Fig. 2) becausethey provide an opportunity to study and compare levels
and patterns of population structurefrom morphologicaland geneticalperspectives. In fact, Kenneth Parkes stated (in Arbib 1981) that "It is quite apparent
that Passerellailiaca must have the most extreme variations in bill size and shape
of any embedhzine,
certainlyin North America and probablyin the world." This
studywasnot intendedas a taxonomic revision, althoughit producedresultswith
taxonomic implications. Individuals were collectedfrom isolatedbreedingpopulations and more-or-lesscontinuouslydistributed habitat, different breeding
habitats,and from regionsin which individualspossess
differentmorphological
characteristics(Fig. 3). The samplingprotocol was designedto allow study of
geographicaland ecologicalcorrelatesof variation, suchas isolationand habitat,
respectively.
Becausevariation is the raw matedhalin the evolutionary process,considerable
attention has been focused on documenting variation in natural populations. I
use quantitative analysesof geneticand morphological variation to describevariation within and among populations.Covadhationof morphologicaland environ-


4

ORNITHOLOGICAL

MONOGRAPHS

NO. 40

)1ogy

FIGURE1. Breedingrangesof the 18 subspecies
of the Fox Sparrowin North America. Subspecific
taxonomy follows the A. O.U. Check-list (1957); seealso Miller (1956). The rangesare generalized,

becausebreedingFox Sparrowsare not continuouslydistributed over the range of each subspecies.
Three subspecies
groupsare recognized(Swarth 1920): Iliaca (iliaca, altivagans,zaboria), Unalaschcensis(unalaschcensis,
insularis,sinuosa,annectens,townsendi,
fuliginosa),and Schistacea
(schistacea,
megarhyncha,stephensLbrevicauda,fulva, canescens,olivacea,swarthL monoensis).Winter range is
the southernUnited States,extremenorthernMexico, and coastalregionsof the westernUnited States.


GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

5

mental features is used to assessthe geography of adaptation. For example, is
phenotypicvariation consistentwith expectationsof ecogeographicrules?Do
characterscovary in similar ways, and do the sexesshow similar patterns of
variation? Assessmentof variation at 38 enzyme loci allows a geneticperspective
on morphologicaltraits, as well as contributinginformation on geneticalpopulation structureand geneflow. How is geneticvariation apportionedin indiviudals
and populationsand among populations?Are isolatedpopulationsrelatively less
variable or genetically differentiated?What is the nature of gene flow? Can a
historical pattern of fragmentation, or evolutionary history, of populations be

discerned?If geographicvariants are a stagein the speciationprocess(Simpson
1953; Rensch 1959; Gotfid and Johnston 1972; Bock 1979; Charlesworth et al.
1982), then the origin of interspecificdifferencesmight be deducedfrom studies
of populations of Fox Sparrows differentiated to various degrees (Vuilleumier
1980). Thus, I compared variation among local populations, subspecies,and
recently evolved species(Zink 1982) to elucidate potential morphological and
genetic correlatesof the speciationprocess.Thus, in these ways I addressthe
evolutionary significanceof geographicvariation in the Fox Sparrow.
PREVIOUS STUDIES OF GEOGRAPHIC VARIATION IN Fox SPARROWS

The Fox Sparrow was the subject of two intensive, broad-scale surveys of
geographicvariation (Swarth 1920; Linsdale 1928). Swarth (1920) studied characteristics of external morphology such as wing and tail lengths, bill size, and
colorationto clarify subspecies
limits. Swarth recognizedthree distinct subspecies
groups:Iliaca, Unalaschcensis,and Schistacea(see legend to Fig. 1). The Iliaca
group, representedby three subspecies,rangesin summer throughout northern
North America, exclusiveof the northwestcoast.Birds are typically reddishwith
relatively short tails and streaked backs. The Unalaschcensisgroup is distributed
along the northwest coast, and birds are typically dark brown in coloration, with
relatively short tails and medium-sized bills, the latter character showinga north
to south clinal increasein size; seven subspeciesare recognized.A notable feature
of this group is the leap-frog pattern of migration wherein subspecieswith the
darkest coloration winter in humid conditions, south of the winter range of the
subspecies
breedingto the southof them. The darker subspecies
breedin relatively
more arid conditions, where one might expect a lighter coloration. Thus, coloration seemsinfluenced by conditions of the winter and not the breeding environment. The third subspeciesgroup, Schistacea, breeds in the mountains of the
western United States,and contains eight subspecies.Members of the Schistacea
group have gray backs with reddish wings and tails, a relatively long tail, and
marked variation in bill size over short geographicdistances.

Swarth concluded that the Iliaca and Unalaschcensisgroupswere most similar
(closelyrelated?),and he developeda historicalscenarioto explain overall patterns
both in the speciesand within each of the three groups.
Linsdale (1928) determined that patterns of geographicvariation in 16 skeletal
charactersparalleled those obtained by Swarth. Linsdale was one of the first to
document concordance between character sets, now termed a test of the "nonspecificityhypothesis" (Sheath and Sokal 1973).
Thus, it has been appreciatedfor over 50 yearsthat extensivegeographicvari-


6

ORNITHOLOGICAL

MONOGRAPHS

NO. 40

ation exists in Fox Sparrows, both in coloration and in skin and skeletal features.
In fact, the number of subspecies,18, ranks third in North America, behind the

Song Sparrow (Melospiza melodia) and Horned Lark (Eremophila alpestris)
(A.O.U. 1957). Becauseof the marked morphologicaldifferentiationamong subspecies,and the monographsby Swarth and Linsdale, it was decidedto undertake
a quantitative description of geneticvariation that could be used to contrast with
morphological patterns of variation, the latter of which was also assessedusing
modern analytical techniques.
STUDY SITES, SAMPLING DESIGN AND TECHNIQUES,
BRIEF

SUMMARY

OF NATURAL

AND

HISTORY

The precise localities of the 31 sample sites are given in the Appendix and
depicted more generally in Figure 2 and Table 1. Site codes,elevation, habitat

type,an estimateof populationdensity,samplesizes,andsubspecific
designation
for the population at each site are also given in Table 1. The general breeding
rangesof the subspeciesare shown in Figure 3; scale drawings of heads of male
Fox Sparrows depict geographicvariation in bill size and shape.
Swarth (1920) and Linsdale (1928) describedthe natural history of Fox Sparrows, and I found their accountsto be highly accurate. Martin (1979) discusses
geographicvariation in song.Aspectsof the breedingdistribution and ecologyof
Fox Sparrows are summarized in Table 2. Fox Sparrows breed in two distinct
habitats in the region I surveyed. In the Great Basin, Fox Sparrows breed in
riparian thicketsconsistingof alder (Alnussp.), water birch (Betula occidentalis),
willows (Salix spp.), Ribes, and other species.These habitats are generallylinearly
distributed along stream courses,becoming somewhat more expansive at canyon
heads. Densities of Fox Sparrows in riparian situations are generally lower than
in chaparral (discussedbeyond). Breeding sites in the Great Basin occur from
about 1,980 m to 3,050 m. Riparian habitats are often disjunct, being separated
by large expansesof uninhabitable (to breeding Fox Sparrows) sagebrushdesert,
an environment typical of much of the Great Basin. Becausemany of thesewater
coursesare fed by springs, I assume that this habitat is available annually to
breedingFox Sparrows,without periodsof local habitat extinction.
West of the Great Basin in the Sierra Nevada, Cascades,North Coast Ranges,
and TransverseRange (Fig. 3), Fox Sparrowsbreed in a very different environment, namely, montane chaparral(seeOrnduff 1974). Thesebrushfields,occurring

from 1,220 m to 3,000 m, include a variety of plant species,most commonly
Arctostaphylospatula, Ceanothusspp., and Castanopsissempervirens.However,
in the Greenhorn Mountains, south of the main Sierra Nevada, Fox Sparrows are
found in elderberry (Sambucus sp.) thickets.
Montane chaparral occursboth on soilsand slopestoo steepor poor in nutrients
for timber and as the natural successionalvegetation on lands deforestedby fires
or logging(Beaver 1976). As a result, montane brush fields vary in age, size, and
vegetation structure. As a post-fire successionalstage, montane chaparral reaches
a density sufficientto supportFox Sparrowswithin approximately 10 years(Bock
and Lynch 1970). For example,at the Cherry Lake (CHER) site,the forestburned
in 1966 and by 1978 the brush was sufficientto support a low density of Fox
Sparrows,which doubtlessimmigrated from nearby breeding sites. Fox Sparrow
densitieschangeas the brush field matures and apparently peak at approximately


GEOGRAPHIC

VARIATION

IN THE FOX SPARROW

7

Passerella

iliaca

o

stephensi

ß

megarhyncha

ß

monaen$i$

0

cane$cen$

ß

brevicauda

A

fulva

ß

$chistacea

'YRA(•)

BLAC(X)

FIGURE

2. Locationof the31 collecting
sites;for precise
locations
seeAppendixI. The subspecific
designationof eachsampleis indicatedby symbols.


8

ORNITHOLOGICAL

MONOGRAPHS

NO. 40

t?t• fulva

P.L$chistaceo
•

t•.• monoen$i$

P,i brevicaudo

.PLcane$cens

P,Lmegorhyncl•o

•,Lstephens/
'• :

FIGURE3. Breedingdistribution of sevensubspecies
of the Fox Sparrowin Oregon,Nevada, and
California. Rangesare generalizedbecausebreedingFox Sparrowsare not continuouslydistributed
within the boundariesof each subspecies.
Scaledrawingsof headsof malesillustrategeographic
variationin bill sizeand approximateshapeonly. Subtleplumagedifferences
shownhereare typical
of individualvariationin all populationsand are not meantto indicatediagnosticgeographic
differences.


GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

TABLE

9

1

DESCRIPTIONOF STUDY SITES(FIG. 2) FOR SAMPLES
OF FOX SPARROWS.
PRECISE

LOCALITY DESCRIPTIONS ARE GIVEN IN APPENDIX

I. SAMPLE SIZES INCLUDE ALL

INDIVIDUALS COLLECTEDAT EACHSITE. IN SOMEANALYSES,NOT ALL SPECIMENS
WERE USED BECAUSE OF DAMAGED CHARACTERS. SUBSPECIFIC TAXONOMY

FOLLOWS
GreNfELL AND MILLER (1944)
Number

of

Elevation

Sil•

Code

Im-

(m)

Males

Females

matins

Habitat • Density:

Subsl:•cies

A.
B.
C.
D.
E.
F.

BERN
PINO
REDM
DOME
LOOK
SHAY

2,210
2,530
1,920
2,410
2,350
1,650

22
18
16
19
17
29

5
8
3
6
8
6

1
0
0
0
I
3

C
C
C
C
C
C

4
4
3
4
4
4

stephensi

stephensi
stephensi
stephensi
stephensi
megarhyncha

G.
H.
I.
J.

MTOM
JACK
CHER
EBET

2,230
2,010
1,550
1,890

11
17
4
13

6
6
5
4

0
3
3
1

C
C
C
C

3
4
1
3

rnegarhyncha
megarhyncha
rnegarhyncha
rnegarhyncha

K.

MONO

2,290

6

3

4

C

2

monoensis

L.
M.
N.
O.
P.
Q.
R.
S.
T.
U.
V.
W.
X.
Y.
Z.
1

WALK
WOOD
TAHW
TAHE

SAGE
BUCK
LASS
SHAS
SPEN
LAUG
WARN
ODEL
BLAC
YOLL
SAWY
PYRA

2,410
1,950
2,010
2,130
1,920
1,650
1,830
1,800
1,220
1,430
1,860
1,580
2,070
1,370
1,650
1,580

8
15
17
9
14
11
34
17
10
11
18
12
15
22
9
12

8
4
5
I
4
2
18
11
3
5
11
4
4

8
8
4

1
5
0
1
3
2
13
2
2
1
3
0
0
8
0
0

C
C
C
C
C
C
C
C
R

C
C
C
C
C
C
C

2
3
4
2
5
2
5
4
2
3
3
3
3
4
2
2

rnegarhyncha
3
megarhyncha
3
rnegarhyncha

rnegarhyncha
rnegarhyncha
megarhyncha
rnegarhyncha
rnegarhyncha
megarhyncha
4
rnegarhyncha
4
fulva
fulva
brevicauda
brevicauda
megarhyncha
rnegarhyncha

2.
3.

WHIT
RUBY

2,680
2,680

22
10

3
2

MART
STEN

2,070
2,230

3
8

3
0
I
I

R
R

4.
5.

8
3
2
3

R
R

1

2

canescens
schistaces
schistaces
fulva

449

176

62

C = montane chaparral;R = riparian.
Densityestimatesare subjectiveestimatesof the densityof breedingFox Sparrowsand the extentof suitablehabitatat eachsite.
A I indicatesthe lowestdensityand a 5 representsa dense,extensivelydistributed,localbreedingcolony.
Samplestreated as intergradesbetweenmonoensisand megarhynchaby Grinnell and Miller (1944).
Essentiallyon the borderofmegarhynchaandfulva.

one pair per hectare (Bock and Lynch 1970; Bock et al. 1978; Savage 1978).
Although the populationsin the TransverseRange (PINO, BERN) are isolated
(Fig. 3), in general the distancebetween suitable chaparral habitats is less than
that between riparian thickets in the Great Basin. Patches of montane chaparral
are sufficientlydenseand widespread to effect a quasi-continuousdistribution in
the Sierra Nevada

and Cascades.

Specimenswere obtained from isolated sites, areas of continuous distribution,
sitesof different elevation, riparian thickets(Great Basin), and chaparralof various

ages(western mountains). General collecting areas were determined from range
maps in Grinnell and Miller (1944) and Miller (1956), and precisecollectingsites


10

ORNITHOLOGICAL

TABLE

MONOGRAPHS

NO. 40

2

ASPECTSOF THE BREEDING DISTRIBUTION AND ECOLOGY OF FOX SPARROWSOF
THE SCHISTACEA SUBSPECIESGROUP

Aspectsof:

Distribution
Elevation
Habitat type
Habitat stability
Breedingdensity
Breedingsite
Bill size

Sierra Nevada, Cascades,

Coast and Transverseranges

Great Basin

Continuousto disjunct
1,200 to 3,000 ra
Montane chaparral
Seralstates(ephemeral)
Low to high
2odimensional
Medium to Large

Disjunct
2,300 to 3,000 ra
Riparian (e.g., willows)
Relatively stable
Usually low
Linear
Small

were chosen while in the field; collecting localities were spaced at about 40 km
intervals, except for a few instanceswhen the sampleswere more closely spaced

for analysisof microgeographicvariation. Some sites were chosento duplicate
samplestaken in the 1920s by Lindsdale (1920); the recent sampleswere analyzed
for temporal variation (Zink 1983). Birds were collectedin June,July, and August
of 1978-1980, using a shotgun or mist-nets. The timing of the collecting effort
insuredthat individuals collectedwould representthe local breedingcommunity.
Precisedates, itineraries, and site descriptionsare on file at the Museum of Vertebrate Zoology (MVZ), University of California, Berkeley, California.
Specimenswere prepared as either study skins plus partial skeletonsor as

complete skeletons. Below, "skin" refers to a standard study skin preparation.
For both types of skeletonpreparations,specimenswere dried (out of sunlight),
and then cleanedby a dermestidbeetle colony. On complete skeletonpreparations,
standard skin measurements(describedbelow) were taken on completely dried
"roughed-out" specimensprior to use of beetles.These "skin" measurementsare
comparableto thosetaken on prepared(and dried) study skins (seeJohnsonet
al. 1984 for further comments on methods).
MATERIALS

AND

METHODS

ELECTROPHORESIS

Within three hours of collection of each specimen, samples of liver, heart,
kidney, and pectoral muscle were frozen in liquid nitrogen. Sampleswere subsequently stored at -76øC until used for electrophoresis.Tissue extracts were
preparedby mincing approximately 0.5 cm3 of tissue(liver and musclecombined;
heart and kidney not used)and combining it with an equal volume of de-ionized
water, and then centrifuging this mixture at 16,000 rpm for 40 min at 4øC. The
supernatant(aqueoustissueextract) was frozen at -76øC and the tissuepellet
discarded.

Gels for horizontal electrophoresiswere made of 12% starchand the appropriate
buffer solution. Electrophoretic conditions for the 38 presumptive genetic loci
examined are given in Table 3. After electrophoresis,the gel was slicedhorizontally
and each slice staineddifferentially usingprotein assaysdescribedby Selanderet
al. (1971) and Harris and Hopkinson (1976). Interpretation of bands on gels

GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

TABLE

11

3

ELECTROPHORETIC CONDITIONS USED FOR STUDY OF FOX SPARROWS
Gel typeI

Volts--hours

Loci2,•

LiOH (2)
TC 8 (5)

300
130

3

4

TM (9)
Poulik (3)

100
250

4
3

liver: LGG; LA-1,2; NP; GDA; LAP; EST-1,4; AB1,2
liver: ICD-I,2; MDH-1,2; Acp (=EAP); PGM-2; LDH-1,2; GPT
muscle:ADA; MPI; GOT-1,2; GPD-I,2; AK
liver: 6-PGD; G-6-PDH; ADH; GDH; SOD-l,2
muscle: GPI; AB-3,4; ACON; CK-1,2

Phos-Cit •

200

3

muscle: GaPDH

Numbers refer to buffer types in Selander et aL (1971).
Abbreviationsfor loci follow Harris and Hopkinson (1976).
Many loci are seorableon several gel types and with several tissues.
Conditions

available from author.

followedHarris and Hopkinson (1976), Barrowcloughand Corbin (1978), and
Avise et al. (1980a). Unless variation was unambiguous,a locuswas not scored.
Stained gels were photographed and saved.
For each population I constructeda table of allelic frequenciesfor each locus.
Measuresof within-populationgeneticvariability were: (1) percentageof loci
polymorphic, calculated as the number of loci with two or more alleles divided

by 38 (POLY99 and POLY95, dependingon whetherthe frequencyof the most
commonallelewas <99% or 95%, respectively),(2) averagenumber of allelesper
polymorphiclocus(NALL), and (3) averageindividualheterozygosity
(H). H was
calculatedby averagingindividual heterozygosities
in each population. That is,
if an individual was heterozygousat three loci, its H estimateis 3/38or 0.079.
Values for eachindividual in a populationsamplewere then averaged(+s.e.).
Also,theexpected
H, Hexp,assuming
Hardy-Weinbergequilibrium,wascalculated
as

• xi?
H=•1i=•1- j=l
where xi is the frequency of the jth allele at the ith locus, k• is the number of

alleles at the ith locus, and N is the total number of loci examined (38). The
varianceof Hexphasa theoreticalexpectationwhichmay differfrom the empirical
s.c. describedabove (Nei 1978; Corbin 1981).
To test the observedfrequencyof genotypesfor departuresfrom Hardy-Weinberg expectations,I followed Barrowclough(1980a) by comparingwith a chisquare test the observed and expected numbers of heterozygotessummed over

all variable loci in a population.The degreesof freedom(d.f.) for this test are the
number of alleles minus one at each locus summed over all loci. Loci at which

only one heterozygotewas expectedwere combined (Lewontin and Felsenstein
1965).
To detect patterns in measuresof within-population geneticvariation, partial
correlationand multiple regressionanalyseswere used.The samplesize,latitude,
longitude, and elevation at each site were coded as independent variables, and
the followingcharacteristics
servedas dependentvariables:H, POLY99, POLY95,
NALL, and the frequencyof the most common allele at the most polymorphic


12

ORNITHOLOGICAL

MONOGRAPHS

NO. 40

loci (LGG, LA-2, EST-D, EAP, ADA, GPI, and NP). The analysis(BMDP6R;
Dixon 1979) controls for correlationsamong the independentvariables, gives
partial correlationcoefficientsamongthe dependentvariables,and assesses
whether or not the independent variables can statistically predict values of the dependent
variables when used in a multiple regressionanalysis.
Population structure was examined with F-statistics, following the methods of
Wright (1978). Three different methodsof computingFsT were used:"Wright's,"
uncorrected, and corrected, the latter of which involves subtraction of an error
term becauseof finite sampling of genesper population. The FsT value calculated

over all variable loci was divided with its empirical s.e. and probably can be
treatedbasa t statistic,with the d.f. equal to the number of loci minus one (see
Barrowclough1980a for a brief descriptionof the terminology).
The geneticdistancemeasuresof Nei (1978) and Rogers (1972) were computed
to measure the degreeof differentiation between populations. These measures,
used extensivelyfor other organisms,permit comparisonsacrosstaxa (seeAvise
and Aquadro 1982). A phenogram,portraying the geographicpattern of genetic
distances,was constructedfrom the matrix of Rogers' D-values; the unweighted
pair-group method usingarithmetic averageswas used(UPGMA; seeSneathand
Sokal 1973). Phenogramsgroup samplesas a function of levels of similarity (i.e.,
low distancesare similar). Theoretically, when using genetic distances,samples
with a common evolutionary history, or those connectedby gene flow, should
clustertogetherif rates of characterstate changeare uniform (Felsenstein1982).
Other methodsexist for constructingbranchingdiagramsfrom distancematrices
(e.g., Farris 1981; Swofford 1981; seeFelsenstein1982 for a review). However,
the genetic distancesin this study are so low that confidencein any branching
structure

is tenuous.

Slatkin (1981) proposeda method to estimate levels of gene flow in natural
populationsusing allelic frequencydata. The simulationsof Slatkin showedthat
the conditional average frequency of an allele [p(i)] is basically independent of
the assumedselectionintensity and mutation rate but dependsheavily on the
overall level ofgene flow. The data required are the averagefrequencyof an allele
conditioned on the number of populations in which it occurs,p(i), and the occupancynumber, L the number of samplesin which the allele wasdetected.Slatkin
then showedthat by plotting p(i) versusi/d (d -- total number of localitiesor
samples),levels of gene flow can be assessedas high, low, or medium. Use of a
recent refinement(Slatkin 1985a) of the 1981 method did not alter conclusions
about gene flow in the Fox Sparrow (Zink, unpubl. data).

MORPHOLOGY

STUDY SKIN MEASUREMENTS

Nine characterswere measuredwith dial calipers(recordedto nearest0.05 mm)
on studyskins,or dried specimensprior to preparationas skeletons:(1) ORETL-lengthof the outerrectrix,measuredfrom point of insertionof the centralrectrices
to tip of the outer rectrix. (2) WINGL--length of the outer primary, measuredas
the chord of the unflattened wing from the bend of the wing to the tip of the
outermostprimary. Excessivewear of the longestprimary preventeduse of this


GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

13

character. (3) HINTL--length of hind toe plus claw, measured from the ventral
base of the hind toe to the tip of the hind claw. Variance due to curvature of the
toe and claw on dried specimens was not sufficient to necessitatedeletion of the
character. (4) TARSL--length of tarsus, measured from the mid-point of the
posteriorsurfacejuncture of the tibiotarsusand the tarsometatarsusto the anterior

lower edgeof the last large scuteon the tibiotarsus,a point consistentlyapparent

on each specimen. (5) BILL-l--length of bill from anterior rim of nares to the
tip of the upper mandible. (6) BILL-2--1ength of lower mandible, from the anteflor-most inner edgeof ramus to the tip of the lower mandible. (7) BILLW-width of lower mandible measuredat its base(widestpoint). (8) BILLD-1 --depth
of bill measuredthrough a plane passingperpendicularthrough the anterior-most
tip of the nostril. (9) BILLD-2--depth of bill measuredfrom the baseof the lower
mandible (the widest point, on lateral aspect)to a point on the culmen directly
above the anterior edge of the nostril.
SKELETAL MEASUREMENTS

Measurements of 15 skeletal charactersjudged relatively accurate (Zink 1983)
were taken on each specimen: (1) SKULW--maximum width of skull acrossthe
bullae, (2) SKULL--partial length of skull measuredfrom suture at posterior end

ofbulla to a notch on the anterior face of the post-orbital process,(3) CORAL-length of coracoid, (4) SCPEW -- width of the proximal end of the scapula, (5)
STERL--length of the sternum, (6) PSYNL -- posterior synsacrum length, (7)
SYNMW--maximum width of the synsacrum,(8) FEPEW--width of the proximal end of the femur, (9) FEDEW--width of the distal end of the femur, (10)
FEMRL--length
of the femur, (11) TIBOL--length of the tibiotarsus, (12)
HTROL--length of the trochanter (humerus), (13) HUMRL--length of the humerus,(14) ULNAL--length of the ulna, and (15) ULPEW--width of the proximal
end of the ulna. Most of these measurements are pictured and described more
fully in Robins and Schnell (1971).
NUMERICAL ANALYSIS OF SKIN AND SKELETAL CHARACTERS

Study skin (SKIN) and skeletal (SKEL) data were analyzed separatelyas were
males and females because of known sexual dimorphism (Linsdale 1928). For
eachpopulation sample,means,variances,and coefficientsof variation (CV) were
computed for each character. Missing values, due to damaged or missing bones,
were replacedby population and sex meansto allow multivariate analysis.However, inserting means weights the population mean, decreasesthe variance, and
increasesthe d.f., all of which are factors that exaggeratedifferencesbetween
groups. Hence, if a specimen lacked more than 2 skin or 3 skeletal characters, it
was excluded from analysisbeyond the calculation of basic univariate statistics.

Analysisof variance(ANOVA) was usedto assess
geographicheterogeneityfor
each character. Inspection of product-moment correlation coefficients,computed
for eachpair of characters(and basedon all individuals) revealedwhich characters
in each data set exhibited similar patterns of variation. Geographic variation for
some characterswas illustrated with pie diagrams. Although no two characters
showedexactlythe samegeographicpattern, inspectionof the characterplots and


14

ORNITHOLOGICAL

MONOGRAPHS

NO.

40

correlation coefficientsbetween pairs of charactersdepicts the geographyof character variation.

To estimate the size component of geographicvariation, correlation coefficients
were computed between sample means for each character and the mean cuberoot of massper site. In addition, a regressionanalysiswas used to determine the
amount of variance in a character that is "explained" by cube-root of mass.
Multivariate analysisof variance (MANOVA) was performed on each data set
to test the hypothesisthat the group centroidsare significantlyheterogeneousin
multivariate space.Principal componentsanalysis(PCA) was used to explore the
generalpatternofphenotypic similarity amongpopulationsamplesin multivariate
morphometric space, and to identify linear combinations of variables that best

summarize charactervariation within and among samples.Raw data were first
log•o transformed. The principal components are orthogonal, unrotated, and

extractedfrom the covariancematrix calculatedover all individuals.To compare
within-sample charactervariation at different sites, 10 of the largestsampleswere
analyzedseparately,and the relative values of character"loadings" inspectedfor
similarity.
Individuals' scoreson the first three principal componentswere analyzed with
the SS-STP method (Gabriel 1964; Gabriel and Sokal 1969; Power (1970) and

Johnson(1980) provideexamplesof this techniquein avian studies).This analysis
delimits a group of population samplesfor a given character,such that addition
of another sample would result in a significantF-value (ANOVA). The resultant
"maximally non-significantsubsets,"computed for each principal component,
are illustrated on a map of localities and provide indications of the geographic
structureof variation in PC scores.Samplesare ranked by PC values from largest
to smallest, rather than by geographicproximity. The scheme used to code pie
diagramsfor charactervariation was usedalsohere to depict locality mean scores
on the first three principal components, portraying a complex pattern of morphologicalvariation in one dimension.
Clusteranalysiswasusedto explorefurther the patternofpheneticrelationships
amongthe samples.The taxonomicdistance(djk)measureand correlationcoefficient, computed from variance-standardized character means for each population, were used to construct an Operational Taxonomic Unit (OTU) by OTU
matrix of distancesor correlation coefficients.The UPGMA and WPGMA (Sneath
and Sokal 1973) algorithms were used on the OTU by OTU matrices. The degree
to which a phenogramrepresentsthe similarity or distancematrix was evaluated
with the cophenetic correlation coefficient(roo).
UPGMA phenogramswere also computed from matrices of taxonomic distances and correlation coefficients that had been constructed from character means

for males that were first divided by the mean cube-root of male massat each site
and then transformed to log•o. This procedure produced groupingsof samples,
perhapslessinfluencedby size,whichmight portraypatternsof variation in shape.

A canonicalcorrelationanalysis(BMDP6M, Dixon 1979) wasperformed,using
samplemeansfor each character(sexesseparate)and the following locality and
climatic variables: elevation (ELEV), latitude (LATI), longitude (LONG), May
mean temperature (MAYT), averagemaximum May temperature (MAYX), average minimum May temperature (MAYM), June mean temperature (JUNT),


GEOGRAPHIC

VARIATION

IN THE

FOX

SPARROW

15

average maximum June temperature (JUNX), average minimum June temperature (JUNM), averageJuly temperature (JULT), averagemaximum July temperature(JULX), averageminimum July temperature(JULM), April precipitation
(APPR), and total annual precipitation (ANNP). The weather data were taken
from recent U.S. Forest Service publications. Most of the weather stations were
located within 20 km of the collecting sites, but often differed in elevation. As a
consequence,temperature values were corrected for elevation by adjusting the
value usedby IøF for every 400-foot differencein elevation betweenthe collecting
site and weather station (Hopkins 1938). The canonicalcorrelation analysistests
for independenceof patterns between two data sets.In this analysis,the objective
is to "explain" morphological patterns of variation in terms of the environmental
data. If independenceis refuted, the analysisindicates which environmental variables are primary determinants of the non-independenceof the two data sets.
RANDOMNESS

IN GEOGRAPHIC

MANTEL

PATTERNS'

TESTS

Mantel's (1967) test comparestwo distancematricesfor congruenceof pattern.
It teststhe hypothesisthat the pattern of distancesin one matrix (dependent)is
independentof the pattern of distancesin the secondmatrix (hypothesis).Here,
two hypothesismatricesare a matrix of minimum geographicdistances(GEOG)
between each pair of sites,and a matrix of the reciprocalsof geographicdistances
(REGE). For dependentmatrices,I usegeneticand morphologicaldistances(males
only). Many alternative hypothesismatrix structuresexist (Sokal 1979). For example, if population sampleswere taken on opposite sidesof a barrier to gene
flow, e.g., a mountain range, the minimum geographicdistance between sites
(acrossthe range) would not be as appropriate as the path distance around the
mountain range,a more biologically realisticgeneflow corridor. Use of reciprocals
of geographicdistance "in effect considerall longer distancesto be about equal
while emphasizing differencesbetween short distances.This procedure increases
the statisticalpowerof the analysisto reveallocalgeographic
patterningwhereas
testsinvolving actual distancesare more usefulin evaluatingregionalgeographic
patterns" (Joneset al. 1980).
If, for example, matrix comparisonsindicated that geneticand geographicdistances were independent, then the pattern of genetic distances might not be a
simple function of isolation by distance(as representedby the particular hypothesismatrix). In this study, GEOG and REGE "hypothesis" matrices were used
becauseI lacked information on patterns of gene flow among these samplesof
Fox Sparrows.Becauseany two matricescan be compared,geneticand morphological matrices were contrastedas well. In addition to t-values resulting from
Mantel's test, matrix correlation coefficientsare alsocomputed to illustrate further
the degreeof matrix association.Douglas and Endlet (1982) and Schnell et al.

(1985) provide further notes on methodology,and Joneset al. (1980) provide a
useful empirical demonstration.
The Mantel procedure generatesa matrix of t-values between distance matrices.
When testingseveralmatrices,Douglasand Endlet (1982) recommendedthat a
corrected t-value be used to reject the null hypothesis of independence of the
matrices.For the presentstudy, nine t-values were computed;thus, the corrected


16

ORNITHOLOGICAL

MONOGRAPHS

NO.

40

probability value for a Type 1 error is 0.05/9, or 0.0056, for which the corresponding t-value is 2.88. Therefore, an observed t-value must exceed 2.88 to
reject the null hypothesis (independence);if a t-value exceeds2.88, then the
matrices share a common structure to some degree.
RESULTS

ELECTROPHORETIC ANALYSIS
LOCUS LEVEL

An initial survey at 38 loci of 150 individuals representingmost collecting sites
indicated that 14 loci were sufficientlypolymorphic to survey for remaining individuals (breeding adults only). Therefore, 14/38 (36.8%) loci were considered
polymorphic, and all calculationsthat follow assume24 loci to be monomorphic
and fixed for the same allele in all 31 population samples.

Barrowcloughand Corbin (1978) and Avise et al. (1980a) describethe electro-

phoreticphenotypes
ofheterozygous
individuals
forseveral
lociin variousspecies
of wood-warblers and thrushes. My results agree with these studies, and with
those documented for the same loci in other vertebrates. Therefore, I refer to
electrophoretic phenotypes (= electromorphs) as alleies.
Notes follow on electrophoretic patterns for loci that are sometimes difficult to
interpret. (1) Esterase-D (EST-D; 4-methylumbelliferyl acetateesterase).The threebanded pattern of heterozygotesat this locus is consistentwith the interpretation

that the active stateof this enzyme is a dimeric condition (Harris and Hopkinson
1976). A fluorescentlamp was used in a darkroom to causethe bands (alleles)to
fluoresce.After scoringgelsfor EST-D, the position of the alleles was marked by
punching a hole in the gel with a stirring rod. Gels were then rinsed with deionized water and treated with a visual esteraseassay (alpha NP + FBRR salt;
Harris and Hopkinson 1976); this caused a complex series of bands to appear.
Other than the most anodal locus (EST-1), the other bands were not interpreted
becauseit seemedthat gene productsof severalloci had similar mobilities, thus
obscuringresolution of singleloci. The visual stain did not detect EST-D, because
none of the colored bands was coincident with the position of bands at EST-D,
thereby implicating single-locuscontrol of EST-D. (2) ErythrocyteAcid Phosphatase (EAP; 4-methylumbelliferyl phosphatase).Presumed heterozygotesat this
locus showed a two-banded phenotype consistentwith the interpretation (Harris
and Hopkinson 1976) that the active form of this enzyme is monomeric. This
locuswas clearly different from the visual acid phosphatase(ACP), becauseheterozygotesat the latter locus exhibited the expectedthree-banded pattern (the
ACP locuswasnot usedhere).(3) Peptidases.Four loci werescoredon LiOH gels,
LAP, LGG, LA-1, and LA-2. No variation was detected at LAP. Using leucylglycyl-glycineas substrate,a locus (LGG) with two-banded heterozygoteswas
observed. Using leucyl-alanine as a substrate,two zones of activity appeared,
representingtwo presumptivegeneticloci (LA- 1 and LA-2). Heterozygotesat LA1 showed a three-banded pattern, and at LA-2 a two-banded pattern, suggesting

dimeric and monomeric statusfor these enzymes, respectively.
Consideringthe 14 polymorphic loci and all (619) individuals (Table 4), the
number of allelesper polymorphic locusranged from two (SOD-1, ICD-1, MPI)
to six (LA-2), and averaged 3.5; the average acrossall loci was 1.9 (including