university of north carolina at wilmington...

302

Journal of Mammalogy, 82(2):302–319, 2001

EVOLUTIONARY GENETICS AND PLEISTOCENE BIOGEOGRAPHYOF NORTH AMERICAN TREE SQUIRRELS (TAMIASCIURUS)

BRIAN S. ARBOGAST,* ROBERT A. BROWNE, AND PETER D. WEIGL

Department of Biology, Wake Forest University, Winston-Salem, NC 27109Present address of BSA: Burke Museum of Natural History and Culture andDepartment of Zoology, University of Washington, Seattle, WA 98195-3010

Nucleotide sequence data from the mitochondrial DNA (mtDNA) cytochrome-b gene andallozymic data were used to infer the evolutionary and biogeographic histories of NewWorld tree squirrels of the genus Tamiasciurus. Phylogenetic analyses of the cytochrome-b data support the existence of 3 mtDNA lineages within Tamiasciurus: a western lineageconsisting of populations of T. douglasii from western British Columbia (Canada), Wash-ington, Oregon, and California, and T. mearnsi from northern Baja California (Mexico); asouthwestern lineage consisting of populations of T. hudsonicus from New Mexico andArizona; and a geographically widespread lineage comprising populations of T. hudsonicusfrom the remainder of the species’ range. Levels of mtDNA sequence variation observedwithin and among populations of Tamiasciurus were small (0–2.4%), suggesting that con-temporary geographic patterns of genetic variation in Tamiasciurus have been establishedrelatively recently (i.e., in the Late Pleistocene). Allozyme analyses also support a closerelationship among extant populations of Tamiasciurus. No fixed allelic differences wereobserved among the 3 recognized species and interspecific genetic distances (Nei’s D) weresubstantially less than those typically observed between sibling species. Although differingfrom the current taxonomy in several respects, geographic patterns of genetic variationobserved within Tamiasciurus are similar to those observed in a variety of North Americanboreal forest taxa and most likely reflect effects of forest fragmentation associated withglacial cycles of the Pleistocene.

Key words: biogeography, mtDNA, phylogeography, Pleistocene, Tamiasciurus

The genus Tamiasciurus contains 3 rec-ognized species, T. hudsonicus, T. doug-lasii, and T. mearnsi (Wilson and Reeder1993), which are distributed throughoutmontane and boreal forest regions of NorthAmerica (Fig. 1). The most widely distrib-uted of the 3 species, T. hudsonicus, has ageographic distribution that extends fromAlaska in the northwest, through the GreatLakes region to the southern Appalachiansin the southeast, and southward along theRocky Mountains into northern Arizona.The geographic distribution of T. douglasiiis limited to western North America (Brit-

* Correspondent: [email protected]

ish Columbia [Canada], Washington,Oregon, and California), primarily west ofthe Cascades and the Sierra Nevada. Thegeographic distributions of T. douglasii andT. hudsonicus are largely exclusive, but the2 species are sympatric in parts of BritishColumbia (Smith 1981), Washington (Ste-vens and Nellis 1974), and Oregon (Hattonand Hoffmann 1979). In general, these 2species can be distinguished on the basis ofpelage coloration (Lindsay 1982). However,individuals exhibiting intermediate pelagecoloration occur within these relatively nar-row areas of sympatry in the Pacific North-west, leading Hall (1981) to suggest that T.

ARBOGAST ET AL.—EVOLUTIONARY GENETICS OF TAMIASCIURUSMay 2001 303

FIG. 1.—Geographic distributions (Hall 1981; Lindsay 1981) and sampling localities for the redsquirrel (Tamiasciurus hudsonicus), Douglas’ squirrel (T. douglasii), and Mearns’ squirrel (T. mearn-si). Numbers correspond to those presented in Appendix I.

hudsonicus and T. douglasii may be con-specific.

Tamiasciurus mearnsi occurs only in theforested highlands of northern Baja Cali-fornia (Lindsay 1981; Yensen and Valdes-Alarcon 1999; Fig. 1). Historically, thesepopulations were considered to be conspe-cific with those of T. douglasii (Hall 1981;Townsend 1897), although Allen (1898)proposed that T. mearnsi be elevated to thespecific level. Recent recognition of T.mearnsi as a distinct species (Wilson andReeder 1993) was based largely on thework of Lindsay (1981), who provided ev-

idence that it is morphologically distinctfrom populations of T. douglasii and T.hudsonicus from western North America.

We used mitochondrial DNA (mtDNA)sequence data to infer geographic patternsof genetic variation within the genus Tam-iasciurus and allozyme data to estimatewithin-species genetic variation andamong-species differentiation. We exam-ined if geographic patterns of genetic vari-ation within the genus were consistent withthe current taxonomy and were congruentwith those observed in other North Amer-ican mammals with similar geographic dis-

304 Vol. 82, No. 2JOURNAL OF MAMMALOGY

tributions. We further examined the role ofPleistocene glacial cycles and associatedforest fragmentation in shaping geographicpatterns of genetic variation within Tamias-ciurus.

MATERIALS AND METHODS

Blood or tissue samples were obtained from68 individuals of Tamiasciurus (Appendix I).Cytochrome-b sequence data were generated for43 individuals (32 T. hudsonicus, 10 T. doug-lasii, and 1 T. mearnsi) and allozyme data weregenerated for 58 (44 T. hudsonicus, 12 T. doug-lasii, and 2 T. mearnsi). Both analyses includedsamples from throughout the geographic distri-bution of each species (Fig. 1). All tissue andblood samples were stored at 2708C before lab-oratory analyses.

Analysis of variation in the mtDNA cyto-chrome-b gene.—Total DNA extractions from ei-ther tissues or blood were conducted using eitherthe phenol–chloroform method (Hillis et al. 1996)or the chelex-solution method (Walsh et al. 1991).A 402–base pair fragment of the cytochrome-bgene of mtDNA was amplified using the poly-merase chain reaction (Mullis and Faloona 1987;Saiki et al. 1988). Amplification and sequencingwere performed with the primers L14724 andH15149 (Irwin et al. 1991) as described by Ar-bogast (1999). Automated sequencing (Wake For-est University School of Medicine, Winston–Sa-lem, North Carolina) was used to sequence theentire 402–base pair fragment of the cytochrome-b gene for each individual. To minimize sequenc-ing errors, both strands were sequenced for eachindividual. Sequences were aligned with Se-quencher software (Gene Codes Corp., Ann Ar-bor, Michigan) and visually inspected for errors.Nucleotide sequences also were translated intoamino acid sequences in the computer programMacClade (version 3.07; Maddison and Maddi-son 1997). All unique cytochrome-b nucleotidesequences were deposited in GenBank under ac-cession numbers AF322945–AF322958.

Phylogenetic analyses and estimates of se-quence divergence were performed using thecomputer program PAUP* (Swofford 1999). Toprovide a basis for comparison with previouslypublished data, sequence divergence for all pos-sible pairwise combinations of unique haplo-types was estimated under the 2-parameter mod-el of Kimura (1980). The computer program

MODELTEST (Posada and Crandall 1998),which uses a series of hierarchical likelihood ra-tio tests to evaluate the fit of 56 nested modelsof nucleotide substitution, was used to determinewhich model provided the best fit to the cyto-chrome-b data. The g1 statistic (Huelsenbeck1991) was used as an index to evaluate if suf-ficient levels of phylogenetic signal were presentin the data.

Parsimony and maximum-likelihood analysesinitially were performed with 3 species of Sciu-rus (S. niger, S. aberti, and S. carolinensis;GenBank Accession numbers U46178, U10177,and U46167, respectively), simultaneously des-ignated as outgroups; Sciurus is the presumptivesister genus to Tamiasciurus (Hafner 1984; Haf-ner et al. 1994). The best-fit model (based onthe hierarchical likelihood ratio tests) was thenused to infer a cytochrome-b gene tree under theoptimality criterion of maximum-likelihood, em-ploying the heuristic search option and the tree-bisection-reconstruction branch-swapping algo-rithm. Maximum-likelihood bootstraps under thebest-fit model consisted of 100 iterations, usingthe fast stepwise-addition-search and strict con-sensus-tree options.

Parsimony analysis was performed with allcharacter state changes weighted equally andalso with transitions down-weighted to reflectthe transition : transversion ratio estimated underthe best-fit model. The branch-and-bound searchoption and tree-bisection-reconstruction branch-swapping algorithms were employed. Nodalsupport was estimated with 1,000 bootstrap rep-licates, using the fast stepwise-addition-searchand strict consensus-tree options. Only uniquehaplotypes were included in bootstrap analyses,and all sites were used as characters. For parsi-mony analysis, consistency and retention indiceswere calculated with all characters included andalso with uninformative characters excluded.

Because the most appropriate outgroups forthis study (Sciurus) are relatively divergent fromTamiasciurus, their inclusion in the phylogeneticanalyses was likely to introduce a high degreeof saturation (Halanych et al. 1999). To addressthat potential problem, maximum-likelihood andparsimony were also used to examine unrootednetworks of the unique haplotypes within Tam-iasciurus. With the exception of excluding theoutgroup taxa, these analyses were conducted inthe same manner as described above. Under theassumption of constant evolutionary rates, mid-


point rooting was used to determine the direc-tion of evolution for those networks. Becausethe midpoint method assumes a molecular clock,a likelihood ratio test was used to test foramong-lineage differences in rate of molecularevolution as described in Huelsenbeck and Ran-nala (1997).

Analysis of allozymic variation.—Tissue orblood samples were analyzed via standard starch-gel electrophoresis (Hillis et al. 1996) with bothcontinuous and discontinuous buffers. Sixteen lociwere scored: a-GPDH-1 (EC: 1.1.1.8), G-6PDH-1(1.1.1.49), IDH-1 (1.1.1.42), LDH-1 (1.1.1.27),PEP-1 (using leu-ala; 3.4.–.–), PGM-1 (5.4.2.2),XDH-1 (1.1.1.204), AAT-1 (2.6.1.1), ACO-1(4.2.1.3), GPI-1 (5.3.1.9), MDH-1 (1.1.1.37),PGDH-1 (1.1.1.44), SDH-1 (1.3.99.1), SOD-1(1.15.1.1), and EST-1 and EST-2 (none specified).Allele frequencies for each locus, percent of locipolymorphic (P; using the 95% criterion), and av-erage individual heterozygosity (H) values werecalculated. Genetic distances were calculated asdescribed by Nei (1972).

RESULTS

Phylogenetic analysis of the mtDNA cy-tochrome-b data.—Fourteen unique haplo-types were observed among the 43 individ-uals of Tamiasciurus examined (AppendixI). Of the 402 base pairs of the cytochrome-b gene examined, 19 were variable withinTamiasciurus. Levels of sequence diver-gence among the haplotypes ranged from0.25% to 2.4% (Table 1). The hierarchicallikelihood ratio tests indicated that the TrNmodel (Tamura and Nei 1993) with among-site rate variation incorporated via a dis-crete gamma distribution (a 5 0.2625) pro-vided the best fit to the cytochrome-b dataset consisting of the 14 unique haplotypesof Tamiasciurus plus the outgroups. The es-timated transition : transversion ratio underthis model was 4.77. When the outgroupswere excluded, the hierarchical likelihoodratio tests indicated that the HKY85 model(Hasegawa et al. 1985) provided the best fitto the data. The estimated transition : trans-version ratio was 5.70 for the reduced dataset. The g1-statistics for the 2 data sets (withand without the outgroups included) were22.33 and 20.807, respectively, indicating

that the data possessed nonrandom phylo-genetic signal (Huelsenbeck 1991).

In the parsimony analysis using outgrouprooting, both weighting schemes (transi-tions and transversions weighted equallyand transversions weighted 5:1 over tran-sitions) recovered the same 3 equally par-simonious trees. The 3 trees differed onlyin the extent to which relationships withinthe western mtDNA clade were resolved.Under the equal-weighting scheme, eachtree was 152 steps in length and had con-sistency and retention indices of 0.875 and0.846, respectively. When uninformativecharacters were excluded, the consistencyindex was 0.873. The 3 trees each had a2ln likelihood score of 1,220.792.

The strict consensus of the 3 most par-simonious trees (Fig. 2A) supported the ex-istence of a distinct ‘‘western’’ mtDNAclade (haplotypes VII–XI) within Tamias-ciurus corresponding to populations of T.douglasii (western British Columbia, Wash-ington, Oregon, and California) and T.mearnsi (Baja California, Mexico). A geo-graphically widespread ‘‘eastern’’ clade(haplotypes I–VI) corresponding to all pop-ulations of T. hudsonicus except those fromNew Mexico and Arizona also was recov-ered. However, bootstrap support for thatclade was weak (,50%). The New Mexicoand Arizona haplotypes (haplotypes XII–XIV) did not form a monophyletic group inthat analysis and were basal to the westernand eastern mtDNA clades.

Maximum-likelihood analysis using out-group rooting and the best-fit TrN 1 gam-ma model (Tamura and Nei 1993) produceda single tree (Fig. 2B) with a 2ln likelihoodscore of 1,201.053. Maximum-likelihoodanalysis also supported the existence of 2distinct mtDNA lineages within Tamiasciu-rus, 1 of which was the same western claderecovered in the parsimony analysis, andthe other a southwestern clade comprised ofthe New Mexico and Arizona haplotypes.Maximum-likelihood analysis did not re-cover a monophyletic eastern mtDNAclade. Rather, haplotypes corresponding to


TA

BL

E1.

—K

imur

a(1

980)

2-pa

ram

eter

dist

ance

s(a

spe

rcen

tage

s)am

ong

the

14un

ique

mtD

NA

hapl

otyp

es(I

–XIV

)ob

serv

edw

ithi

nT

amia

s-ci

urus

.H

aplo

type

desi

gnat

ions

are

give

nin

App

endi

xI,

and

phyl

ogen

etic

rela

tion

ship

sam

ong

the

hapl

otyp

esar

ede

pict

edin

Fig

s.2

and

3.D

ista

nces

betw

een

the

Tam

iasc

iuru

sha

plot

ypes

and

indi

vidu

als

repr

esen

ting

3sp

ecie

sfr

omth

epu

tati

vesi

ster

genu

s,Sc

iuru

s,ar

esh

own

for

com

pari

son.

Hap

loty

pe

Hap

loty

pe

III

III

IVV

VI

VII

VII

IX

IX

XI

XII

XII

IX

IV

I II III

IV V VI

0.49

90.

249

0.52

21.

005

0.75

2

0.24

90.

514

1.00

30.

750

0.25

70.

752

0.50

01.

051

0.78

30.

249

VII

VII

IIX X X

IX

II

1.77

22.

031

1.76

82.

026

2.03

11.

772

1.76

82.

026

1.76

52.

022

2.02

61.

768

1.51

51.

772

1.51

21.

768

1.77

21.

515

1.58

61.

855

1.58

21.

584

2.12

01.

852

1.26

01.

515

1.25

71.

512

1.51

51.

260

1.00

51.

260

1.00

31.

257

1.26

01.

005

0.24

90.

249

0.24

90.

752

1.51

5

0.49

90.

499

1.00

51.

772

0.44

90.

750

1.51

21.

003

1.76

81.

772

XII

IX

IVS.

caro

line

nsis

S.ab

erti

S.ni

ger

2.03

12.

290

17.4

820

.66

20.8

7

2.02

62.

285

17.4

520

.27

21.1

9

1.77

22.

031

17.1

520

.31

21.2

3

2.12

62.

395

16.6

820

.99

21.6

9

1.51

51.

772

17.8

120

.31

21.9

4

1.26

01.

515

17.4

820

.66

21.5

8

1.77

22.

031

17.8

122

.11

22.6

7

2.03

12.

290

18.1

522

.48

23.0

4

1.76

82.

026

18.1

222

.06

22.6

3

2.02

62.

285

23.0

022

.43

18.1

2

2.03

12.

290

23.0

421

.74

18.8

2

0.24

90.

500

21.4

220

.66

17.1

5

0.24

921

.58

20.3

116

.82

21.2

319

.96

16.5

0


FIG. 2.—Strict consensus of A) the 3 most parsimonious trees and B) maximum-likelihood treefor the 14 haplotypes (I–XIV) of Tamiasciurus using outgroup rooting. The 3 most parsimonioustrees were identical except for the degree of resolution within the western clade. The maximum-likelihood tree is based on the best-fit model of nucleotide substitution for these data (TrN 1 gamma).Locality information and haplotype designations are given in Appendix I. Bootstrap values .50%(based on 1,000 replicates for parsimony and 100 replicates for maximum-likelihood) are shown ateach node. Bars are used to illustrate the degree of congruence between the cytochrome-b gene treeand the current taxonomy of Tamiasciurus.

the eastern mtDNA lineage recovered in theparsimony analysis (haplotypes I–VI) werelargely unresolved and basal to the 2 recov-ered clades.

Under the best-fit model, the likelihoodratio test of a molecular clock indicated thatno significant differences occurred in therate of cytochrome-b evolution among the14 haplotypes of Tamiasciurus (2ln likeli-hood without molecular clock enforced 5682.131; with molecular clock enforced 5687.820; d.f. 5 12; P 5 0.30). Thus, thecritical assumption required for using themidpoint method to root the haplotype net-work of Tamiasciurus (that the data areevolving according to a molecular clock)seems to have been met for those data.

Both weighting schemes used in the mid-point-rooted parsimony analysis (transitionsand transversions weighted equally andtransversions weighted 6:1 over transitions)produced the same 3 trees, which differedonly in the extent to which relationshipswithin the western mtDNA clade were re-solved. The 3 trees were 20 steps each and

had a 2ln likelihood score of 682.13. Con-sistency and retention indices (both withand without uninformative characters ex-cluded) were 1.0 for those trees. Maximum-likelihood analysis produced a single treethat was identical in topology to the strictconsensus parsimony tree. The maximum-likelihood tree had a 2ln likelihood of682.11. Thus, when outgroups were exclud-ed and the midpoint method was used toroot the network consisting of the 14unique haplotypes of Tamiasciurus, parsi-mony and maximum-likelihood convergedon the identical tree topology (Fig. 3). Bothmethods recovered the same 3 majormtDNA clades within Tamiasciurus: awidespread eastern clade (haplotypes I–VI),a western clade (haplotypes VII–XI), and asouthwestern clade (haplotypes XII–XIV).Although a sister relationship between east-ern and western clades was recovered inboth analyses, bootstrap support for that re-lationship is weak (,50%). Parsimony andmaximum-likelihood bootstrap support forthe 3 clades was generally high (.85%),


FIG. 3.—Phylogeny recovered by both parsimony and maximum-likelihood for the 14 unique hap-lotypes (I–XIV) of Tamiasciurus using midpoint rooting. Locality information and haplotype desig-nations are given in Appendix I. Parsimony bootstrap values .50% (based on 1,000 replicates) areshown above each node, and maximum-likelihood bootstrap values .50% (based on 100 replicates)are shown below each node. Bars are used to illustrate the degree of congruence between the cyto-chrome-b gene tree and the current taxonomy of Tamiasciurus.

except for the node that included the easternOregon haplotypes (haplotypes V and VI)within the eastern clade, where bootstrapsupport was about 65%. The monophyly ofthe remaining haplotypes corresponding tothe eastern clade was well supported.

Allozymic analysis.—Of the 16 loci ex-amined, 8 were nonvariable: EST-1, a-GPDH-1, G-6PDH-1, IDH-1, LDH-1, PEP-1 (using leu-ala), PGM-1, and XDH-1.Based on the 8 variable loci, average indi-vidual heterozygosity was 6.7% and 14.6%for T. hudsonicus and T. douglasii, respec-tively, and the percentage of loci polymor-phic was 37.5% for both species (Table 2).Because allozymic data were available fromonly 2 individuals of T. mearnsi, estimates

of heterozygosity and polymorphism forthat species were not meaningful and allelefrequencies were not interpreted to be re-alistic. Rather, those values were listed onlyfor comparison with the other 2 species todetermine whether alleles were shared orunique. Genetic distance between T. hud-sonicus and T. douglasii was estimated tobe 0.015. Because of small sample size, noestimate was made using T. mearnsi.

DISCUSSION

Genetic variability and population struc-ture of Tamiasciurus.—Perhaps the moststriking result of this study is the relativelysmall amount of genetic variability foundwithin and among populations of Tamias-


TA

BL

E2.

—A

llel

icfr

eque

ncie

sat

8va

riab

lelo

cian

des

tim

ates

ofav

erag

ein

divi

dual

hete

rozy

gosi

ty(H

)an

dpe

rcen

tage

oflo

cipo

lym

orph

ic(P

)fo

rT

amia

sciu

rus.

Loc

i

Spe

cies

AA

T-1

12

3

AC

O-1

12

ES

T-2

12

34

GP

I-1

12

MD

H-1

12

PD

GH

-1

12

3

SD

H-1

12

SO

D-1

12

H (%)

P (%)

T.h

udso

nicu

s(n

544

)T

.do

ugla

sii

(n5

12)

T.

mea

rnsi

(n5

2)

0.75

0.39

1.00

0.25

0.56

0.05

0.05

0.08

0.95

0.92

1.00

0.04

0.50

0.38

0.18

0.50

0.57

0.36

0.01

0.46

0.14

0.12

0.86

0.88

1.00

0.06

0.04

0.50

0.94

0.96

0.50

0.01

0.50

0.16

0.17

0.50

0.83

0.83

0.89

0.92

1.00

0.11

0.08

0.99

1.00

1.00

0.01

6.7

14.6

37.5

37.5 ciurus. For example, levels of estimated

mean sequence divergence in the mtDNAcytochrome-b gene among the 3 recognizedspecies of Tamiasciurus are 1.0–2.4%, orabout one-third of those observed withinthe similarly distributed northern flyingsquirrel (Glaucomys sabrinus—Arbogast1999; Demboski et al. 1998). Based on theallozyme analyses, the genetic distance val-ue of 0.015 observed between T. hudsoni-cus and T. douglasii also is well below thattypically observed between sibling species(Avise 1994). Although the sample for T.mearnsi is too small for definitive conclu-sions, the data also are consistent with ahigh degree of genetic similarity between itand the other 2 species. Overall, allele fre-quencies, heterozygosity, and polymor-phism values (Table 2) are typical of thosefound in a large outbreeding population(Smith and Wayne 1996).

Although a variety of problems are as-sociated with estimating evolutionary datesfrom estimates of molecular divergence(Arbogast and Slowinski 1998; Hillis et al.1996), levels of mtDNA and allozymic var-iability suggest that the origin of presentgeographic structuring within Tamiasciurusis a recent phenomenon. Based on a gam-ma-corrected rate of cytochrome-b diver-gence for primates of about 5–6%/millionyears (Arbogast and Slowinski 1998) and arelative rate of cytochrome-b evolution forrodents about 1.5–2 times faster than thatof primates (Britten 1986; Dewalt et al.1993; Li et al. 1987, 1990; Wu and Li1985), a gamma-corrected rate of cyto-chrome-b evolution for rodents is estimatedto be about 7.5–12%/million years. Use ofthis rate and minimum values of estimatedpairwise sequence divergence between the3 major clades of Tamiasciurus, whichrange from 1.03% to 1.77% (Table 1), pro-duces mean estimated dates of cytochrome-b gene divergence of about 80,000–240,000years ago. Because rates of molecular evo-lution are lineage specific and because thesedates represent only mean estimates and arelikely to have large associated standard er-


rors (Arbogast and Slowinski 1998; Hilliset al. 1996), they should be viewed withcaution. However, the relatively small lev-els of mtDNA and allozymic differentiationobserved suggest that the majority of geo-graphic structuring of genetic variationwithin Tamiasciurus probably occurred inthe Late Pleistocene. Furthermore, giventhat persistence of ancestral polymorphismsis predicted to lead to estimated dates ofgene divergence that predate episodes ofactual population separation and speciation(potentially by hundreds of thousands ofyears—Edwards 1997), divergence of pop-ulations corresponding to the 3 majormtDNA clades of Tamiasciurus may haveoccurred as recently as the Wisconsin gla-ciation or Wisconsin–Holocene transition.

Phylogenetic analysis of the cytochrome-b data.—The relatively few variable char-acters within Tamiasciurus (consisting pri-marily of 3rd-position transitions), coupledwith the availability of only relatively dis-tant outgroups, made phylogenetic analysisof the Tamiasciurus sequence data chal-lenging. In particular, difficulty was en-countered in rooting the phylogeny of Tam-iasciurus. This problem is manifested in thefailure of the outgroup analyses to resolvethe more basal relationships within Tamias-ciurus and in bootstrap values that are gen-erally low (Fig. 2).

To address the problem of having onlyrelatively divergent taxa with which to rootthe phylogeny, we reconstructed unrootednetworks of the 14 unique haplotypes ofTamiasciurus (Fig. 3). Those networkswere then rooted using the midpoint meth-od. Several lines of evidence led us to favorthis topology as being that which best re-flects evolutionary relationships withinTamiasciurus. First, theoretical work hasshown that the probabilities of obtaining thecorrect rooted tree are substantially lowerthan the probabilities of obtaining the cor-rect unrooted tree (Sourdis and Krimbas1987). Second, because the outgroup taxaare so divergent from the ingroup taxa(about an order of magnitude greater than

those within Tamiasciurus in terms of cy-tochrome-b sequence divergence; Table 1),inclusion of the outgroups is likely to adda large amount of saturation, homoplasy,and potentially long-branch attraction intothe analyses (Halanych et al. 1999; Hendyand Penny 1989; Maddison et al. 1992; Mi-yamoto and Boyle 1989; Smith et al. 1992;Wheeler 1990). Third, when the outgroupsare excluded parsimony and maximum-likelihood methods recover the same topol-ogy (Fig. 3), suggesting that this result isrobust across a variety of phylogenetic as-sumptions (Halanych et al. 1999). Finally,when the outgroups are excluded, bootstrapvalues increase at all major nodes withinthe phylogeny of Tamiasciurus, and consis-tency and retention indices increase.

Biogeographic implications.—Lindsay(1987) proposed that Pleistocene forestfragmentation played an important role indetermining patterns of speciation and geo-graphic variation within Tamiasciurus. Thegenetic data presented here support Pleis-tocene forest dynamics as an important fac-tor in shaping the biogeographic and evo-lutionary history of the group. For example,the close relationship of T. mearnsi fromBaja California with populations of T.douglasii to the north supports Lindsay’s(1981) biogeographic argument for a LatePleistocene separation of these 2 popula-tions. Lindsay (1981) proposed that themechanism responsible for the isolation ofT. mearnsi on the Baja Peninsula was thefragmentation of appropriate forest habitatfor Tamiasciurus during Pleistocene inter-glacials. To the north and west of the SierraSan Pedro Martir, California chaparral(Brown 1994) is currently a barrier betweenpopulations of T. mearnsi and T. douglasiiin the Sierra Nevada (Yensen and Valdes-Alarcon 1999). However, during Pleisto-cene glacial periods, life zones were 600–1,000 m lower than at present, forming acorridor of coniferous forest between theSierra San Pedro Martir and the Sierra Ne-vada (Lindsay 1981; Savage 1960). Ances-tors of T. mearnsi could have invaded the


FIG. 4.—Geographic distributions of eastern (gray), western (white), and southwestern (black)mitochondrial DNA clades of Tamiasciurus based on midpoint-rooted parsimony and maximum-likelihood trees (Fig. 3). Open circles indicate collection localities for individuals from which mtDNAcytochrome-b data were obtained. Complete locality information is given in Appendix I.

Baja Peninsula during 1 or more of the me-sic cycles that began about 700,000 yearsago (Hafner and Riddle 1997). T. mearnsilikely would have become isolated from T.douglasii most recently about 12,000 yearsago (Lindsay 1981; Yensen and Valdes-Alarcon 1999).

Analysis of the mtDNA data also revealsthe presence of a distinct lineage of Tam-iasciurus in the southern Rocky Mountains(Figs. 2–4). Lindsay (1987) noted morpho-logic differences in the southwestern pop-ulations of Tamiasciurus and suggestedthose differences were the result of stronginfluences imposed by the vicariant bioge-

ography of the southwestern United Statesand factors related to Allen’s rule (an eco-geographic relationship between tempera-ture and extremity length). Historical frag-mentation of coniferous forests in thesouthern Rocky Mountains almost certainlyisolated southwestern populations of Tam-iasciurus from more northern populations atvarious times in the Pleistocene (Harris1990; Lindsay 1987), most recently about6,000–9,000 years ago (Hafner and Sulli-van 1995). During the Wisconsin glacialmaximum, the combination of the ColoradoRiver drainage and continuous montaneglaciers south of the river would have ef-


fectively divided the boreal forest habitat ofnorthern Colorado from that of New Mex-ico and Arizona (Hafner and Sullivan1995). As a result, populations of Tamias-ciurus persisting in the Southwest wouldhave become isolated from those in thenorthern Rockies. Subsequently, as the Wis-consin glacial period came to a close andthe climate warmed, populations in NewMexico and Arizona would have becomemore fragmented, isolated from one anotheron montane islands separated by lowlandareas of nonboreal habitat.

Evidence for the historical isolation ofconiferous forests in the southern RockyMountains is supported by molecular dataof Douglas-fir (Pseudotsuga menziesii),which has been shown to possess a pro-nounced phylogeographic discontinuity inthe southwestern United Sates (Aagaard etal. 1995). This discontinuity within Doug-las-fir seems to be highly congruent withthe southwestern mtDNA discontinuity ob-served in Tamiasciurus. Similar phylogeo-graphic discontinuities observed in theAmerican pika (Ochotona princeps—Haf-ner and Sullivan 1995), the long-tailed vole(Microtus longicaudus—Conroy and Cook2000), and the tassel-eared squirrel (Sciurusaberti—Lamb et al. 1997), also may havebeen produced or reinforced by similar his-torical fragmentation of boreal forest andhigh-altitude montane habitats during theLate Pleistocene.

The southeastern United States and thenorthern Rocky Mountains are the mostlikely locations for forest refugia for pop-ulations corresponding to the easternmtDNA clade of Tamiasciurus during themost recent glacial maximum. Existence ofPleistocene boreal forest refugia in thesoutheastern United States is supported byfloristic reconstructions of the region and avariety of mammalian fossils (Davis 1983;Kurten and Anderson 1980; Ritchie 1987).However, because fossil material fromTamiasciurus dating to the Wisconsin gla-cial maximum has not been found furthersouth than Pennsylvania (Kurten and An-

derson 1980), the extent of the distributionof Tamiasciurus in the southeastern UnitedStates during this period is unclear. ThemtDNA haplotypes of Tamiasciurus foundonly in the Blue Mountains of easternOregon (haplotypes V and VI) may havepersisted during the most recent glaciationin 1 or more boreal refugia in the northernRocky Mountains. Molecular evidencefrom several plant taxa from western NorthAmerica suggests that this area may havebeen an important refugia during glacialmaxima (Soltis et al. 1997).

Despite having an extensive geographicdistribution (Fig. 4), the eastern mtDNAclade of Tamiasciurus possesses littlemtDNA variation (Table 1). A similar pat-tern has been observed in a variety of taxa(Arbogast 1999; Avise et al. 1984, 1987;Gill et al. 1993; Green et al. 1996; Zink1996; Zink and Dittmann 1993) that arepresumed to have undergone rapid, postgla-cial, northward dispersal from a limitedpopulation source (Ball and Avise 1992;Gaines et al. 1997; Gill et al. 1993). Paly-nologic evidence indicates that after themost recent glacial retreat, boreal forestsquickly recolonized most of northern NorthAmerica, with spruce (Picea) populationspreviously confined to the southeasternUnited States recolonizing Alaska by about8,000 years ago (Ritchie 1987). A rapidnorthward and westward recolonization ofboreal forest such as this would have pro-vided an opportunity for a similarly rapidrecolonization of northern North Americaby southeastern populations of Tamiasciu-rus and the eventual coalescence of thesepopulations with those that had persisted inthe northern Rocky Mountains.

The most likely locations of Late Pleis-tocene refugia for the southwestern andwestern clades of Tamiasciurus are theAmerican Southwest and the Pacific coastalregion of the United States, respectively. Incontrast to the eastern mtDNA lineage ofTamiasciurus, which seems to have under-gone an extensive postglacial range expan-sion, the southwestern and western mtDNA


lineages of Tamiasciurus appear to have ex-panded their ranges only slightly northwardafter glacial retreats. Asymmetrical patternsof postglacial range expansion and second-ary contact between populations previouslyisolated in separate refugia may explaingeographic locations of major phylogeo-graphic discontinuities observed in Tamias-ciurus (Fig. 4).

The phylogeographic discontinuity ob-served in Tamiasciurus in the PacificNorthwest (Fig. 4) aligns closely with ge-netic and distributional discontinuities in avariety of boreally distributed vertebrates.These include the mule deer (Odocoileushemionus—Cronin 1992), black bear (Ur-sus americanus—Byun et al. 1997; Croninet al. 1991; Demboski et al. 2000; Woodingand Ward 1997), marten (Martes american-us) and dusky shrew (Sorex monticolus—Demboski et al. 2000), northern flyingsquirrel (G. sabrinus—Arbogast 1999);leopard frog (Rana pretiosa—Green et al.1996), giant salamanders (Dicamptodon—Good 1989), common yellowthroat (Geoth-lypus trichas) and rufous-sided towhee (Pi-pilo erythrophthalmus—Ball and Avise1992), chickadee (Parus—Gill et al. 1993),and fox sparrow (Passerella iliaca—Zink1996). A particularly high level of concor-dance seems to exist between the phylogeo-graphic patterns of Tamiasciurus and G. sa-brinus (Arbogast 1999). The major differ-ences between the phylogeographic patternsof these 2 taxa are that the western clade ofTamiasciurus seems to extend slightly fur-ther to the east than does that of the westernclade of G. sabrinus and that the currentdistribution of G. sabrinus does not extendsouthward into regions of the United Satesoccupied by the southwestern clade of Tam-iasciurus. The genetic discontinuity sharedby Tamiasciurus and a wide variety of bo-really distributed vertebrates in the PacificNorthwest supports an evolutionary scenar-io wherein at least 2 previously separatedboreal forest biotas have come into recentsecondary contact in this region (Arbogast

1999) forming a pronounced biogeographic‘‘suture zone’’ (Remington 1968:322).

Taxonomic implications.—The nature ofspecies boundaries within Tamiasciurus hasremained somewhat enigmatic to date,largely because of an incomplete under-standing of whether the recognized speciesare reproductively isolated. For example,Smith (1968) and Stevens and Nellis (1974)collected specimens with intermediate pel-age coloration from British Columbia andWashington, respectively, which they pro-posed to be hybrids. Intermediate pelagealso has been observed in the Blue Moun-tains of Oregon (Hatton and Hoffmann1979) and on Vancouver Island (Lindsay1982), suggesting that hybridization alsomay occur in these areas. To address taxo-nomic problems posed by the existence ofintermediately colored squirrels, Lindsay(1982) evaluated the morphology of 20 cra-nial characters in populations of T. doug-lasii and T. hudsonicus from the PacificNorthwest. He concluded that the 2 speciesare morphologically distinct and that squir-rels exhibiting intermediate coloration con-sistently clustered with individuals of 1 spe-cies or the other. Thus, Lindsay (1982) ar-gued that T. douglasii and T. hudsonicusare isolated reproductively and representseparate species. However, Verts and Car-raway (1998) found a large number of Tam-iasciurus from central Oregon that couldnot be classified as either T. douglasii or T.hudsonicus based on the characters used byLindsay (1982). The presence of uniquemtDNA haplotypes (haplotypes V and VI)in this region indicates that populations ofTamiasciurus currently inhabiting the BlueMountains of north-central and northeasternOregon have become genetically differen-tiated from both western and more easternpopulations of Tamiasciurus, although ap-parently to a lesser degree in the latter case.This may explain difficulties in classifyingindividuals from this region as either T.hudsonicus or T. douglasii (Verts and Car-raway 1998).

Because mtDNA typically is inherited in


a matrilineal fashion in mammals (Avise etal. 1987), the cytochrome-b data presentedhere bear only circumstantially on speciesboundaries within Tamiasciurus. However,from a taxonomic perspective, it is impor-tant that the mtDNA and allozymic dataagree in suggesting that all populations ofTamiasciurus examined in this study areclosely interrelated. Furthermore, absenceof fixed allelic differences among the 3 rec-ognized species of Tamiasciurus suggeststhat they may not be isolated reproductive-ly. Although samples from the Baja Cali-fornia populations of Tamiasciurus (cur-rently assigned to T. mearnsi) are small,phylogenetic analysis of the mtDNA se-quence data consistently included T. mearn-si from Baja California within the westernclade along with individuals assigned to T.douglasii (Figs. 2 and 3). Bootstrap valuessupporting this relationship are high, andlevels of sequence divergence between T.mearnsi and T. douglasii are quite low (Ta-ble 1). Thus, Hall’s (1981) inclusion of T.mearnsi as a subspecies within T. douglasiiseems to be a more accurate reflection ofspecies boundaries within the genus than isthe current taxonomy (Lindsay 1981; Wil-son and Reeder 1993).

The mtDNA data also support the pres-ence of a distinct southwestern clade withinTamiasciurus (Fig. 3). Populations of Tam-iasciurus in the southwestern Rocky Moun-tains show sufficient morphologic distinct-ness from more northern populations of T.hudsonicus to have warranted the earlierspecific status of T. fremonti before the dis-covery of apparent introgression betweenthe 2 groups (Hardy 1950). Recent mor-phologic studies also support the presenceof a distinct group of Tamiasciurus in thesouthwestern United States (Lindsay 1987).As such, the combined mtDNA and mor-phologic evidence suggest that T. hudsoni-cus is comprised of 2 distinct evolutionarylineages.

With the exception of those populationscomprising the southwestern clade, 3 of the4 methods of phylogenetic reconstruction,

including the phylogeny presented in Fig.3, support the presence of a widespreadclade consisting of the remaining popula-tions of T. hudsonicus (Figs. 2–4); the ex-ception (Fig. 2B) most likely reflects prob-lems associated with the inclusion of distantoutgroups in the phylogenetic analyses. Thegeographic boundary between the easternclade and the western T. douglasii–T.mearnsi clade seems to be coincident withthe John Day River in Oregon. This resultalso is supported by mtDNA sequence datafrom the more rapidly evolving control re-gion (B. S. Arbogast, in litt.). Analysis ofvocalization data also suggests that the JohnDay River may be an important geographicboundary between populations of Tamias-ciurus. C. C. Smith communicated to Hat-ton and Hoffmann (1979) that squirrelssouth and west of the John Day producedcalls characteristic of T. douglasii, whereasthose east of the river gave calls character-istic of T. hudsonicus (Verts and Carraway1998). Thus, evidence exists that 2 previ-ously separated, but closely related, evolu-tionary lineages of Tamiasciurus come intocontact, or nearly into contact, in this re-gion.

Our results do not support the currenttaxonomy for Tamiasciurus. The lack ofobserved genetic differentiation of the BajaCalifornia populations (currently recog-nized as T. mearnsi) suggests that thesepopulations do not represent a distinct spe-cies. Furthermore, the low levels of geneticdifferentiation between T. douglasii and T.hudsonicus, coupled with the existence ofapparent hybrids and individuals that can-not be classified to species based on mor-phology where the 2 species come into con-tact in the Pacific Northwest (Verts andCarraway 1998), suggest that T. douglasiiand T. hudsonicus are probably conspecific.Thus, recognition of a single, phenotypi-cally variable species, T. hudsonicus, com-prised of 3 subspecies, T. h. hudsonicus(eastern clade), T. h. douglasii (westernclade, including T. mearnsi), and T. h. mo-gollonensis (southwestern clade), would


best reflect the currently available geneticand morphologic data. Alternatively, eachof the 3 proposed subspecies could be rec-ognized as separate phylogenetic species(Cracraft 1983). Although differing fromthe current taxonomy in several respects,geographic patterns of genetic variation ob-served within Tamiasciurus are consistentwith those observed in a variety of NorthAmerican boreal forest taxa. These com-mon phylogeographic patterns most likelywere generated as a result of changes in thedistribution of boreal forests associatedwith glacial–interglacial cycles of the Qua-ternary.

ACKNOWLEDGMENTS

We thank the following scientists and insti-tutions for providing material for this study: J.A. Cook, University of Alaska Museum, Fair-banks; D. J. Hafner, New Mexico Museum ofNatural History; G. J. Kenagy and S. A. Rohwer,Burke Museum of Natural History and Culture,University of Washington; P. Myers, the Muse-um of Zoology, University of Michigan; S. G.Mech, Washington Sate University; D. S. Rog-ers, Brigham Young University; B. R. Riddle,University of Nevada, Las Vegas; P. D. Sudman,Tarleton State University; G. M. Wilson,Oklahoma State University; and T. L. Yates andW. L. Gannon, Museum of Southwestern Biol-ogy, University of New Mexico. We also thankW. Conner, G. J. Kenagy, K. A. Kron, A. J.Schwandt-Arbogast, and S. Vignieri for helpfulcomments on earlier drafts of this manuscript,and K. A. Kron for use of laboratory facilities.This project was funded in part by a Grant-in-Aid of Research from the American Society ofMammalogists and a Theodore Roosevelt Me-morial Grant from the American Museum ofNatural History to B. S. Arbogast.

LITERATURE CITED

AAGAARD, J. E., S. S. VOLLMER, F. C. SORENSON, AND

H. STRAUSS. 1995. Mitochondrial DNA productsamong RAPD profiles are frequent and strongly dif-ferentiated between races of Douglas-fir. MolecularEcology 4:441–447.

ALLEN, J. A. 1898. Revision of the chickarees, orNorth American red squirrels (subgenus Tamiasciu-rus). Bulletin of the American Museum of NaturalHistory 10:249–298.

ARBOGAST, B. S. 1999. Mitochondrial DNA phylo-geography of the New World flying squirrels (Glau-

comys): implications for Pleistocene biogeography.Journal of Mammalogy 80:142–155.

ARBOGAST, B. S., AND J. B. SLOWINSKI. 1998. Pleisto-cene speciation and the mitochondrial DNA clock.Science 282:1955a.

AVISE, J. C. 1994. Molecular markers, natural historyand evolution. Chapman and Hall, New York.

AVISE, J. C., J. E. NEIGL, AND J. ARNOLD. 1984. De-mographic influences on mitochondrial DNA line-age survivorship in animal populations. Journal ofMolecular Evolution 20:99–105.

AVISE, J. C., ET AL. 1987. Intraspecific phylogeogra-phy: the mitochondrial DNA bridge between popu-lations genetics and systematics. Annual Review ofEcology and Systematics 18:489–522.

BALL, R. M., JR., AND J. C. AVISE. 1992. MitochondrialDNA phylogeographic differentiation among avianpopulations and the evolutionary significance ofsubspecies. The Auk 109:626–636.

BRITTEN, R. J. 1986. Rates of DNA sequence evolutiondiffer between taxonomic groups. Science 231:1393–1398.

BROWN, D. E. (ED.). 1994. Biotic communities: south-western United States and northwestern Mexico.University of Utah Press, Salt Lake City.

BYUN, S. A., B. F. KOOP, AND T. E. REICHMEN. 1997.North American black bear mtDNA phylogeogra-phy: implications for morphology and the HaidaGwaii glacial refugium controversy. Evolution 51:1647–1653.

CONROY, C. J., AND J. A. COOK. 2000. Phylogeographyof a post-glacial colonizer: Microtus longicaudus(Rodentia: Muridae). Molecular Ecology 9:165–175.

CRACRAFT, J. 1983. Species concepts and speciationanalysis. Pp. 159–187 in Current ornithology (R. F.Johnston, ed.). Plenum Press, New York.

CRONIN, M. A. 1992. Intraspecific variation in mito-chondrial DNA of North American cervids. Journalof Mammalogy 73:70–82.

CRONIN, M. A., S. C. AMSTRUP, AND G. W. GARNER.1991. Interspecific and intraspecific mitochondrialDNA variation in North American bears (Ursus).Canadian Journal of Zoology 69:2985–2992.

DAVIS, M. B. 1983. Quaternary history of deciduousforests of eastern North America and Europe. An-nals of the Missouri Botanical Garden 70:550–563.

DEMBOSKI J. R., B. K. JACOBSEN, AND J. A. COOK.1998. Implications of cytochrome b sequence vari-ation for biogeography and conservation of thenorthern flying squirrels (Glaucomys sabrinus) ofthe Alexander Archipelago, Alaska. Canadian Jour-nal of Zoology 76:1771–1777.

DEMBOSKI, J. R., K. D. STONE, AND J. A. COOK. 2000.Further perspectives on the Haida Gwaii glacial re-fugium. Evolution 53:2008–2012.

DEWALT, T. S., P. D. SUDMAN, M. S. HAFNER, AND S.K. DAVIS. 1993. Phylogenetic relationships of pocketgophers (Cratogeomys and Pappogeomys) based onmitochondrial DNA cytochrome b sequences. Mo-lecular Phylogenetics and Evolution 2:193–204.

EDWARDS, S. V. 1997. Relevance of microevolutionaryprocesses to higher level molecular systematics. Pp.251–278 in Avian molecular systematics and evo-lution (D. P. Mindell, ed.). Academic Press, NewYork.


GAINES, M. S., J. E. DIFFENDORFER, R. H. TAMARIN,AND T. S. WHITTAM. 1997. The effects of habitatfragmentation on the genetic structure of smallmammal populations. Journal of Heredity 88:294–304.

GILL, F. B., A. M. MOSTROM, AND A. L. MACK. 1993.Speciation in North American chickadees: I. Patternsof mtDNA genetic divergence. Evolution 47:195–212.

GOOD, D. A. 1989. Hybridization and cryptic speciesin Dicamptodon (Caudata: Dicamptodontidae). Evo-lution 43:728–744.

GREEN, D. M., T. F. SHARBEL, J. KEARSLEY, AND H.KAISER. 1996. Postglacial range fluctuation, geneticsubdivision and speciation in the western NorthAmerican spotted frog complex, Rana pretiosa.Evolution 50:374–390.

HAFNER, D. J. 1984. Evolutionary relationships of theNearctic Sciuridae. Pp. 3–23 in The biology ofground-dwelling squirrels (J. O. Murie and G. R.Michener, eds.). University of Nebraska Press, Lin-coln.

HAFNER, D. J., AND B. R. RIDDLE. 1997. Biogeographyof Baja California peninsular desert mammals. Pp.39–68 in Life among the muses: papers in honor ofJames S. Findley (T. L. Yates, W. L. Gannon, andD. E. Wilson, eds.). Special Publication, The Mu-seum of Southwestern Biology 3:1–290.

HAFNER, D. J., AND R. M. SULLIVAN. 1995. Historicaland ecological biogeography of Nearctic pikas (La-gomorpha: Ochotonidae). Journal of Mammalogy76:302–321.

HAFNER, M. S., L. J. BARKLEY, AND J. M. CHUPASKO.1994. Evolutionary genetics of New World treesquirrels (tribe Sciurini). Journal of Mammalogy 75:102–109.

HALANYCH, K. M., J. R. DEMBOSKI, B. JANSEN VAN

VUUREN, D. R. KLEIN, AND J. A. COOK. 1999. Cy-tochrome b phylogeny of North American hares andjackrabbits (Lepus, Lagomorpha) and the effects ofsaturation in outgroup taxa. Molecular Phylogeneticsand Evolution 11:213–221.

HALL, E. R. 1981. The mammals of North America.2nd ed. John Wiley & Sons, New York 1:1–600.

HARDY, R. 1950. A new tree squirrel from centralUtah. Proceedings of the Biological Society ofWashington 63:13–14.

HARRIS, A. H. 1990. Fossil evidence bearing on south-western mammalian biogeography. Journal of Mam-malogy 71:219–229.

HASEGAWA, M., H. KISHINO, AND T. YANO. 1985. Dat-ing of the human–ape splitting by a molecular clockof mitochondrial DNA. Journal of Molecular Evo-lution 22:32–38.

HATTON, L. E., JR., AND R. S. HOFFMANN. 1979. Thedistribution of red squirrels (Tamiasciurus) in east-ern Oregon. The Murrelet 60:23–25.

HENDY, M. D., AND D. PENNY. 1989. A framework forthe quantitative study of evolutionary trees. System-atic Zoology 38:297–309.

HILLIS, D. M., C. MORITZ, AND B. K. MABLE (EDS.).1996. Molecular systematics. 2nd ed. Sinauer As-sociates, Inc., Publishers, Sunderland, Massachu-setts.

HUELSENBECK, J. 1991. Tree-length distribution skew-

ness: an indicator of phylogenetic information. Sys-tematic Biology 40:257–270.

HUELSENBECK, J., AND B. RANNALA. 1997. Phylogeneticmethods come of age: testing hypotheses in an evo-lutionary context. Science 276:227–232.

IRWIN, D. M., T. D. KOCHER, AND A. C. WILSON. 1991.Evolution of the cytochrome b gene of mammals.Journal of Molecular Evolution 32:128–144.

KIMURA, M. 1980. A simple model for estimating evo-lutionary rates of base substitutions through com-parative studies of nucleotide sequences. Journal ofMolecular Evolution 16:111–120.

KURTEN, B., AND E. ANDERSON. 1980. Pleistocenemammals of North America. Columbia UniversityPress, New York.

LAMB, T., T. R. JONES, AND P. J. WETTSTEIN. 1997. Evo-lutionary genetics and phylogeography of tassel-eared squirrels (Sciurus aberti). Journal of Mam-malogy 78:117–133.

LI, W.-H., G. MANALO, P. M. SHARP, C. O’HUIGIN, AND

Y.-W. YANG. 1990. Molecular phylogeny of Roden-tia, Lagomorpha, Primates, Artiodactyla, and Car-nivora and molecular clocks. Proceedings of the Na-tional Academy of Sciences of the United States ofAmerica 87:6703–6707.

LI, W.-H., M. TANIMURA, AND P. M. SHARP. 1987. Anevaluation of the molecular clock hypothesis usingmammalian DNA sequences. Journal of MolecularEvolution 25:330–342.

LINDSAY, S. L. 1981. Taxonomic and biogeographicrelationships of Baja California chickarees (Tamias-ciurus). Journal of Mammalogy 62:673–682.

LINDSAY, S. L. 1982. Systematic relationship of para-patric tree squirrel species (Tamiasciurus) in the Pa-cific Northwest. Canadian Journal of Zoology 60:2149–2156.

LINDSAY, S. L. 1987. Geographic size and non-sizevariation in Rocky Mountain Tamiasciurus hudson-icus: significance in relation to Allen’s rule and vi-cariant biogeography. Journal of Mammalogy 69:39–48.

MADDISON, D. R., M. RVOLO, AND D. L. SWOFFORD.1992. Geographic origins of human mitochondrialDNA: phylogenetic evidence from control region se-quences. Systematic Biology 41:111–124.

MADDISON, W. P., AND D. R. MADDISON. 1997. Mac-Clade, version 3.07. Distributed by Sinauer Associ-ates, Inc., Publishers, Sunderland, Massachusetts (ondisk).

MIYAMOTO, M. M., AND S. M. BOYLE. 1989. The po-tential importance of mitochondrial DNA sequencedata to eutherian mammal phylogeny. Pp. 437–450in The hierarchy of life (B. Fernholm, K. Bremer,and H. Jornvall, eds.). Elsevier, Amsterdam, TheNetherlands.

MULLIS, K. B., AND F. A. FALOONA. 1987. Specific syn-thesis of DNA in vitro via a polymerase catalyzedchain reaction. Methods in Enzymology 155:335–350.

NEI, M. 1972. Genetic distance between populations.The American Naturalist 106:283–292.

POSADA, D., AND K. A. CRANDALL. 1998. MODEL-TEST: testing the model of DNA substitution. Bioin-formatics 14:817–818.

REMINGTON, C. L. 1968. Suture-zones of hybrid inter-


action between recently joined biotas. EvolutionaryBiology 2:321–428.

RITCHIE, J. C. 1987. Past and present vegetation of thefar northwest of Canada. University of TorontoPress, Toronto, Ontario, Canada.

SAIKI, R. K., ET AL. 1988. Primer-directed enzymaticamplification of DNA with a thermostable DNApolymerase. Science 239:487–491.

SAVAGE, J. M. 1960. The biogeography of Baja Cali-fornia and adjacent seas: evolution of a peninsularherpetofauna. Systematic Zoology 9:184–212.

SMITH, A. B., B. LAFAY, AND R. CHRISTEN. 1992. Acomparison of morphological and molecular rates ofevolution: 28S ribosomal RNA in echinoids. Philo-sophical Transactions of the Royal Society of Lon-don B 338:365–382.

SMITH, C. C. 1968. The adaptive nature of social or-ganization in the genus of tree squirrels Tamiasciu-rus. Ecological Monographs 38:31–63.

SMITH, C. C. 1981. The indivisible niche of Tamias-ciurus: an example of nonpartitioning of resources.Ecological Monographs 51:343–363.

SMITH, T. B., AND R. K. WAYNE. 1996. Molecular ge-netic approaches in conservation. Oxford UniversityPress, New York.

SOLTIS, D. E., M. A. GITZENDANNER, D. D. STRENGE,AND P. M. SOLTIS. 1997. Chloroplast DNA intraspe-cific phylogeography of plants from the PacificNorthwest of North America. Plant Systematics andEvolution 206:353–373.

SOURDIS, J., AND C. KRIMBAS. 1987. Accuracy of phy-logenetic trees estimated from DNA sequence data.Molecular Biology and Evolution 4:159–166.

STEVENS, W. F., AND C. H. NELLIS. 1974. Notes on thehybridization of three mammals in the Ross LakeBasin: squirrel, deer, and deermouse. Appendix H inBiotic survey of Ross Lake Basin, report for January1973–April 1974 (R. D. Taber, ed.). College of For-est Resources, University of Washington, Seattle.

SWOFFORD, D. L. 1999. PAUP*: phylogenetic analysisusing parsimony (*and other methods), version 4.0.Sinauer Associates, Inc., Publishers, Sunderland,Massachusetts (on disk).

TAMURA, K., AND M. NEI. 1993. Estimation of the

number of nucleotide substitutions in the control re-gion of mitochondrial DNA in humans and chim-panzees. Molecular Biology and Evolution 10:512–526.

TOWNSEND, C. H. 1897. Descriptions of a new eaglefrom Alaska and a new squirrel from Lower Cali-fornia. Proceedings of the Biological Society ofWashington 11:145–146.

VERTS, B. J., AND L. N. CARRAWAY. 1998. Land mam-mals of Oregon. University of California Press,Berkeley.

WALSH, P. S., D. A. METZGER, AND R. HIGUCHI. 1991.Chelex 100 as a medium for simple extraction ofDNA for PCR-based typing from forensic material.Biotechniques 10:506–513.

WHEELER, W. C. 1990. Nucleic acid sequence phylog-eny and random outgroups. Cladistics 6:363–368.

WILSON, D. E., AND D. M. REEDER (EDS.). 1993. Mam-mal species of the world. Smithsonian InstitutionPress, Washington, D.C.

WOODING, S., AND R. WARD. 1997. Phylogeographyand Pleistocene evolution in the North Americanblack bear. Molecular Biology and Evolution 14:1096–1105.

WU, C.-I., AND W.-H. LI. 1985. Evidence for higherrates of nucleotide substitution in rodents than inman. Proceedings of the National Academy of Sci-ences of the United States of America 82:1741–1745.

YENSEN, E., AND M. VALDES-ALARCON. 1999. FamilySciuridae. Pp. 239–320 in Mamıferos del Noroestede Mexico (S. T. Alvarez-Castaneda and J. L. Patton,eds.). Centro de Investigaciones Biologicas del No-roeste, La Paz, Mexico.

ZINK, R. M. 1996. Comparative phylogeography ofNorth American birds. Evolution 50:308–317.

ZINK, R. M., AND D. L. DITTMANN. 1993. Gene flow,refugia, and the evolution of geographic variation inthe song sparrow (Melospiza melodia). Evolution47:717–729.

Submitted 20 March 2000. Accepted 3 August 2000.

Associate Editor was Meredith J. Hamilton.


APPENDIX ISampling localities, type of data collected, and mitochondrial (mtDNA) haplotype designations for

individuals of Tamiasciurus hudsonicus, T. douglasii, and T. mearnsi examined in this study. As-signments of species is based on distributional data presented in Hall (1981) and Lindsay (1981).Localities are depicted in Fig. 1. ‘‘X’’ indicates type of data (mtDNA, allozymic, or both) collectedfor a particular individual. Roman numerals designate which of the 14 unique mtDNA haplotypes(I–XIV) was observed in a particular individual.

Taxon State or province County

Museumtissue, GenBank

accession numbera

Allo-zymes mtDNA

mtDNAhaplotypedesigna-

tion

T. hudsonicus TennesseeWest VirginiaOhioMaine

CarterRandolphSummitHancock

BSA226BSA225MSB:NK63048MSB:NK62126

XX

XX

X

III

IIMichiganMichiganMinnesotaMinnesota

MackinacAlgerStationStation

UMMZ:162427UMMZ:167068MSB:NK63189MSB:NK63201

XXXX

XXX

IIIIIIIII

South DakotaColoradoWyomingWyomingWyomingWyoming

LawrenceJacksonTetonTetonBig HornBig Horn

PDS477MSB:NK56591PDS463PDS464PDS474PDS476

XXX

XX

XXXX

IVIIIIII

WyomingWyomingUtahUtahUtahUtah

AlbanyAlbanySan JuanWasatchWasatchWasatch

PDS483PDS484DSR3073DSR3077DSR3078DSR3079

XXXXXX

X

XX

II

IVIV

UtahUtahNew MexicoNew MexicoArizona

WasatchWasatchSandovalSandovalApache

DSR3080DSR3081NMMNH:1767NMMNH:1768MSB:NK17871

XXXXX

XXXXX

IIII

XIIIXIVXII

ArizonaWashingtonWashingtonWashington

ApacheOkanoganOkanoganOkanogan

MSB:NK17872UWBM:BSA181UWBM:BSA184MSB:NK8541

XXXX

XX

XIIII

WashingtonWashingtonWashingtonWashingtonWashingtonAlaska

OkanoganOkanoganOkanoganPend OreilleChelanFairbanksb

MSB:NK8542MSB:NK8560MSB:NK8561SGM4MSB:NK8571AF17840

XXX

XX

X

II

IIIAlaskaAlaskaAlaskaOregonOregonOregon

Hill Islandb

——UnionUnionUnion

AF17873MSB:NK20629MSB:NK21987UWBM:BSA201UWBM:BSA202UWBM:BSA203

XXXXX

XX

XX

IIIIII

VV

OregonOregonOregonOregon

UnionUnionGrantGrant

UWBM:BSA204UWBM:BSA205UWBM:BSA206UWBM:BSA207

XXXX

X

X

V

V


APPENDIX I—Continued.

Taxon State or province County

Museumtissue, GenBank

accession numbera

Allo-zymes mtDNA

mtDNAhaplotypedesigna-

tion

OregonOregonOregonOregon

GrantGrantGrantGrant

UWBM:BSA208UWBM:BSA209UWBM:BSA210UWBM:BSA211

XXXX

XXX

VV

VI

T. douglasiiOregonOregonOregonOregonOregon

GrantGrantGrantGrantKlamath

UWBM:BSA212UWBM:BSA213UWBM:BSA214UWBM:BSA215UWBM:BSA216

XXXXX

XX

X

VIVII

VIIOregonOregonOregonOregonOregonBritish Columbia

KlamathKlamathKlamathLaneLaneVancouverb

UWBM:BSA217UWBM:BSA218UWBM:BSA219UWBM:BSA220UWBM:BSA221BSA:BC423

XXXXX

XXXXXX

VIIVIIVII

VIIIVIII

IX

T. mearnsi

British ColumbiaCaliforniaCaliforniaCaliforniaCaliforniaMexico

Vancouverb

AlpineCalaverasCalaverasHumboldtBaja Californiab

BSA:BC3062MSB:NK8586MSB:NK8582MSB:NK8583MSB:NK8579BSA235

XXX

X

XX

IX

XXI

MexicoMexico

Baja Californiab

Baja Californiab

MSB:NK8062MSB:NK8068

XX

a Tissues and corresponding voucher specimens are housed in the following institutions: University of Alaska Museum, Fairbanks(AF); Museum of Zoology, University of Michigan, Ann Arbor (UMMZ); Museum of Southwestern Biology, University of NewMexico, Albuquerque (MSB); New Mexico Museum of Natural History, Albuquerque (NMMNH); and Burke Museum of NaturalHistory and Culture, University of Washington, Seattle (UWBM). Tissues provided by private individuals (S. G. Mech, B. R.Riddle, D. S. Rogers, and P. D. Sudman) are designated by their initials or the catalog number of the senior author (BSA).

b Noncounty locality.

university of north carolina at wilmington...

Documents