HERPETOLOGICA - University of Idahojacks/Nielsonetal06.pdf · United States and southwestern Canada. These frogs display a number of morpholog-ical adaptations to their unique environments

HERPETOLOGICA

VOL. 62 September 2006 No. 3

Herpetologica, 62(3), 2006, 235–258

E 2006 by The Herpetologists’ League, Inc.

ALLOZYME AND MITOCHONDRIAL DNA VARIATION IN THETAILED FROG (ANURA: ASCAPHUS): THE INFLUENCE OF

GEOGRAPHY AND GENE FLOW

MARILYN NIELSON1,5,7, KIRK LOHMAN

1,6, CHARLES H. DAUGHERTY2, FRED W. ALLENDORF

3,KATHY L. KNUDSEN

3, AND JACK SULLIVAN4

1Department of Fish and Wildlife Resources, University of Idaho, Moscow, Idaho 83844–11362School of Biological Sciences, Victoria University of Wellington, P.O. Box 600, Wellington, New Zealand

3Division of Biological Sciences, University of Montana, Missoula, MT 598124Department of Biological Sciences, University of Idaho, Box 443051, Moscow, ID 83844–3051

ABSTRACT: In this paper we support the division of the previously monotypic genus Ascaphus into twospecies based on an analysis of 23 allozyme loci from 34 populations. We use maximum likelihood to estimatethe Ascaphus phylogeny from the allozyme data and find strongly supported monophyletic Rocky Mountainand Pacific clades. In a nonhierarchal, model-based cluster analysis of the data, each of the 1085 individualgenotypes is correctly assigned to either Ascaphus montanus or A. truei with a high probability. We also finda virtually fixed difference between the species at the Pgm-2 locus. Within A. truei, we find a lack ofsignificant pairwise FST values among populations from the Coast and central Cascades Mountains,suggesting relatively recent range expansion or contemporary gene flow among these populations. OlympicMountains populations form a discrete clade in the allozyme topology and are fixed for a unique allele at theLap locus. These populations remain isolated from the remainder of the species’ range based on pairwise FST

values. The four southernmost A. truei populations each show significant allelic divergence from theremaining populations (based on pairwise FST values), suggesting climate-induced isolation. In addition. weextend mtDNA sampling within the Rocky Mountains and sequence 530 nucleotides from the mtDNACytochrome b (cyt b) gene in 12 previously unsampled A. montanus populations. This additional samplingdefines the geographic extent of a southern mtDNA clade distinguished on average by 0.024 substitutions persite from the northern clade. We use nested clade analysis, a coalescent-based divergence by isolation withmigration model (MDIV), maximum-likelihood estimation of the mtDNA topology, and Bayesian model-based genotype assignment, to test predictions from two hypotheses: the western refugia hypothesis, whichclaims that A. montanus persisted in refugia west of the Snake River during Pleistocene glacial maxima, andthe dual refugia hypothesis, which asserts that A. montanus occupied refugia within the Salmon andClearwater River Valleys during glacial maxima. Our data do not support predictions of the western refugiahypothesis. Nested-clade analysis, estimated dates of lineage divergence supplied by MDIV, and the mtDNAtopology support predictions of the dual refugia hypothesis; however, the allozyme topology fails to supportsome of these predictions. Allozyme and mtDNA data endorse the preliminary recognition of a minimumof two Evolutionarily Significant Units (ESUs) within A. truei: (1) populations from the Olympic Mountainsand (2) populations south of the Umpqua River. Two ESUs are also suggested within A. montanus:(1) populations south of the South Fork of the Salmon River, and (2) populations to the north and west ofthe Salmon River (including the Blue, Wallowa and Seven Devils Mountains). The MDIV analysis indicatesan exchange rate of 10 migrants per generation between the northern ESU and the southern ESU.

Key words: Amphibians; Biogeography; Conservation; Genetic structure; Pacific Northwest; Pacifictailed frog; Rocky Mountain tailed frog

5 PRESENT ADDRESS: 109 Marlene Street, Marshall, MN 562586 PRESENT ADDRESS: Upper Midwest Environmental Sciences Center, 2630 Fanta Reed Road, La Crosse, WI 546037 CORRESPONDENCE: e-mail, [email protected]

235

TAILED frogs inhabit forested headwaterstreams with clean cobble substrates andtemperatures below 21 C (Adams and Frissell,2001) in the mountains of the northwesternUnited States and southwestern Canada.These frogs display a number of morpholog-ical adaptations to their unique environments.Both tadpoles and adults are ventrally flat-tened, tadpoles possess suctorial mouthpartsthat allow them to maneuver in high velocitycurrents, and unlike most anurans, fertiliza-tion in tailed frogs is internal and maleascaphids have a copulatory organ.

Mitochondrial DNA (mtDNA) sequencesreveal considerable variation within Ascaphus(Nielson et al., 2001). The largest divergence(7.8 to 9.1% uncorrected sequence diver-gence) distinguishes Rocky Mountain popula-tions from Coastal and Cascades Mountainpopulations. Nielson et al. (2001) proposedthat this divergence dates to the rise of theCascades Mountain Range and the ensuingrain shadow that currently bifurcates thetailed frogs’ range. Based on mtDNA di-vergence, Nielson et al. (2001) proposed thatRocky Mountain populations represent a dis-tinct species, Ascaphus montanus, a designa-tion that is generally accepted (e.g., Frost,2004).

Morphological studies of the genus (Paukenand Metter, 1971) originally suggested thatAscaphus colonized the northern RockyMountains via the highlands of central andeastern Oregon at the close of the Pleistocene.However, the ancient vicariance hypothesisthat Nielson et al. (2001) invoked to explainthe depth of mtDNA divergence withinAscaphus has recently been substantiated bystatistical tests within the lineage, as well asfor other amphibians with similar distributions(Carstens et al., 2005). The ancient vicariancehypothesis predicts that genetic structure—originating from historic climatic events—willbe found within A. montanus, constrastingwith a lack of significant structure expectedif colonization of the Rocky Mountainsoccurred at the close of the Pleistocene(Brunsfeld et al., 2001). Two potential corol-laries to the ancient vicariance hypothesisexplain contemporary genetic structure interms of the alternative locations of ancientglacial refugia.

The first hypothesis (western refugia hy-pothesis) posits that habitats west of the SnakeRiver served as a refugium for A. montanusthrough the Pleistocene while the body oftheir current range in the northern RockyMountains was glaciated. Under this hypoth-esis, populations from west of the Snake Riverserved as a source for colonization of theremainder of the current range. Predictions ofthis hypothesis include: (1) a grouping ofhaplotypes from the Blue and WallowaMountains at the origin of the A. montanusphylogeny, (2) higher mean heterozygositywithin Blue and Wallowa Mountains popula-tions, and (3) evidence of range expansionfrom west of the Snake River.

The second hypothesis (dual refugia hy-pothesis) is suggested by recent phylogeo-graphic studies of Plethodon vandykei (Car-stens et al., 2004a) and Dicamptodonaterrimus (Carstens et al., 2004b), whichindicate that the Clearwater River and SouthFork Salmon River valleys, respectively,served as refugia for these species duringPleistocene glaciation. These salamander spe-cies have ecological tolerances similar to thatof the tailed frog; therefore, we propose thatboth refugia were used by A. montanus.Predictions under this hypothesis include:(1) the presence of two monophyletic cladeswithin A. montanus, (2) geographic associationof these clades with the purported refugia,(3) evidence of range expansion from tworefugia, and (4) an estimated divergence timeseparating these clades that dates to thePleistocene.

In this paper, we have four objectives. First,we present an extensive allozyme data setfrom across the range of Ascaphus to evaluatethe taxonomic hypothesis proposed by Nielsonet al. (2001) using independent nuclear data.Second, we examine genetic structure withinA. truei using the allozyme data. Third, we testthe western refugia and dual refugia hypoth-eses using mtDNA samples from 12 popula-tions not examined previously, as well asallozyme samples from 14 localities withinthe northern Rocky Mountains. Finally, weintegrate the information from the mtDNAand allozyme data sets to provide a phyloge-netic framework for conservation and man-agement.

236 HERPETOLOGICA [Vol. 62, No. 3

METHODS

Allozymes

Sample collection and scoring.—We exam-ined allozymes in 1085 individuals from 34localities; 14 localities were within the rangeof A. montanus and 20 were within the rangeof A. truei (Table 1; Fig. 1; Appendix I).Allozyme sample collection took place be-tween 1973–1976 prior to the rise of concernover conservation of the tailed frog. Allsampled individuals were larvae, with theexception of several adults from population36 that were sacrificed for other aspects ofDaugherty’s (1979) dissertation research.Many of the sampled larvae would not beexpected to survive to maturity, and thissampling is not known to have negativelyaffected the populations in question. Twenty-three loci were appraised using three buffersystems (Appendix II). Details of collectionand assessment of electromorph homology aregiven in Daugherty (1979).

Phylogenetic analysis.—To estimate thepopulation phylogeny from allozyme data, weused the CONTML program in PHYLIP. Thisprogram implements restricted maximum-likelihood estimation under a Brownian mo-tion model of allele frequency change de-signed to approximate independent geneticdrift among loci (Felsenstein, 1981). The Guslocus was not resolved in the Dollar Creekpopulation (pop. 42); we therefore estimatedtopologies both with the Dollar Creek popu-lation but without the Gus locus and withoutthe Dollar Creek population but with the Guslocus. Bootstrap estimates of nodal support foreach phylogeny were calculated using 200replicate data sets. The CONTML program isunable to analyze data sets with identicalpopulations, and such populations are anunavoidable outcome given a reasonable num-ber of replicate data sets (J. Felsenstein,personal communication). Therefore, we usedthe bootstrap data sets to calculate distancematrices (Cavalli-Sforza’s chord measure;Cavalli-Sforza and Edwards, 1967) in thePhylip program GENDIST, and generated200 neighbor-joining trees using the programNEIGHBOR. Topologies were rooted at themidpoint because a suitable outgroup is notclear. Tailed frogs are considered sister to allother anuran lineages and appear highly

divergent based on several ancestral charac-teristics retained in Ascaphus that have beenlost in all other anuran lineages (Cannatellaand Hillis, 1993; Ford and Cannatella, 1993;but see Hay et al., 1995).

Population level analyses.—We employedanalysis of molecular variance (AMOVA;Cockerham, 1969, 1973) to examine structurein the allozyme data using Arlequin (Schnei-der et al., 2000). We first classified popula-tions as A. montanus or A. truei based oncollection locality and then calculated pairwiseFST values for all populations and tested thesignificance of divergence among populations;pairwise FST values were not corrected formultiple comparisons. Arlequin was also usedto test for deviations from Hardy–Weinbergequilibrium (Guo and Thompson, 1992) andto calculate mean heterozygosity.

mtDNA

Collection and sequencing.—We examinedmtDNA in 60 larvae from 12 populationswithin the range of A. montanus (Appendix I;Fig. 1). There was no overlap between theselocalities and the sampling of Nielson et al.(2001). We extracted DNA using the samemethods described in Nielson et al. (2001; i.e.,Puregene Kit with subsequent cleaning onglass beads).

We initially amplified and sequenced theCytochrome b (cyt b) gene using primersL14115 and SUV as detailed in Nielson et al.(2001); however, we were unable to producehigh quality sequences using the SUV primerin samples from the southeastern portion ofthe species’ range (pops. 12a, 12b and 24–32).We replaced this primer with another Asca-phus specific primer (929V 59-ATGGAAAGC-GAAAAATCGTG-39) that amplified 530 nu-cleotides. We precipitated double strandedPCR products, prepared cycle sequencingreactions and sequenced the cyt b gene usingthe methods described in Nielson et al.(2001). Sequences are deposited in theGenBank database under accession numbersDQ087511–DQ087517.

Phylogenetic analyses.—We aligned andedited sequences using Sequencher (Gene-Codes, Inc.) and conducted all mtDNAphylogenetic analyses using PAUP* (version4.0; Swofford, 1998). For these analyses, we

September 2006] HERPETOLOGICA 237

TA

BL

E1.

—A

llele

freq

uen

cies

and

aver

age

het

eroz

ygos

ity

for

14lo

ciin

34p

opu

lati

ons

acro

ssth

era

nge

ofA

scap

hu

s.P

opu

lati

onn

um

ber

sco

rres

pon

dw

ith

thos

ein

Ap

pen

dix

Ian

dF

igu

res

1an

d2.

Tw

enty

-th

ree

loci

wer

eor

igin

ally

exam

ined

;n

ine

show

edn

ova

riat

ion

inth

ese.

A.

mon

tanu

sA

.tr

uei

Nor

ther

nR

ocky

Mts

.S

alm

onR

Mts

.7

Dev

ilsW

allo

wa

Mts

.B

lue

Mts

.N

orth

ern

Cas

cad

esM

ts.

Cen

tral

Cas

cad

esM

ts.

Coa

stM

ts.

Oly

mp

icM

ts.

Sou

ther

nC

asca

des

Mts

.S

iski

you

Mts

.

Pop

ula

tion

3743

3641

533

3242

3439

4035

1438

5946

2348

4544

4947

5817

5218

5051

5354

5556

1557

Nu

mb

erof

ind

ivid

ual

s

4044

2940

4020

4040

4011

560

4022

4040

4011

754

1540

646

6143

519

4040

4021

406

Loc

us

(alle

le)

Aat

-1a

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

990

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

b—

——

——

——

——

——

——

——

——

——

——

——

0.01

——

——

——

——

——

Aat

-2a

1.00

00.

990

1.00

00.

990

0.75

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

360

0.40

00.

430

0.17

00.

120

0.00

00.

010

0.00

00.

000

0.50

00.

980

0.99

00.

740

0.99

01.

000

b—

0.01

0—

0.01

00.

250

——

——

——

——

——

——

——

—0.

830

0.60

00.

830

0.87

00.

260

1.00

00.

970

1.00

00.

970

0.03

00.

000

0.58

00.

010

—

c—

——

——

——

——

——

——

——

——

——

——

——

0.01

0—

0.02

0—

0.03

0—

—0.

000

——

—

d—

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—0.

010

——

—

Agp

-1a

1.00

01.

000

0.98

00.

980

0.90

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

820

0.93

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

670

b—

—0.

020

0.03

00.

100

——

——

——

——

——

——

0.18

00.

070

——

——

——

——

——

——

——

0.33

0

Gus

a1.

000

0.99

01.

000

0.90

01.

000

1.00

01.

000

ND

1.00

0—

—0.

020

——

0.99

01.

000

1.00

0—

0.29

0—

—0.

090

—0.

040

——

——

——

——

——

b—

0.01

0—

0.10

0—

——

ND

—1.

000

1.00

00.

980

1.00

01.

000

0.01

0—

—1.

000

0.71

01.

000

1.00

00.

910

1.00

00.

960

0.99

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

c—

——

——

——

ND

——

——

——

——

——

——

——

——

0.01

0—

——

——

——

——

Idh-

1a

1.00

01.

000

1.00

01.

000

0.96

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

——

——

——

—0.

010

0.08

00.

650

1.00

01.

000

1.00

01.

000

0.85

00.

440

0.84

00.

600

1.00

01.

000

b—

——

—0.

040

——

——

——

——

—1.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

0.99

00.

920

0.35

0—

——

—0.

150

0.56

00.

160

0.40

0—

—

Idh-

2a

1.00

00.

930

0.93

00.

910

0.84

01.

000

1.00

00.

880

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

880

0.87

01.

000

0.75

00.

870

0.38

00.

190

—0.

630

0.88

00.

680

0.23

00.

930

——

b—

0.07

00.

070

0.09

00.

160

——

0.13

0—

——

——

——

——

——

0.12

00.

130

—0.

250

0.13

00.

630

0.81

01.

000

0.37

00.

130

0.33

00.

780

0.07

01.

000

1.00

0

Lap

a1.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

0—

——

—1.

000

1.00

01.

000

1.00

01.

000

1.00

0

b—

——

——

——

——

——

——

——

——

——

——

——

—1.

000

1.00

01.

000

1.00

0—

——

——

—

Ldh

-2a

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

930

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

b—

——

——

——

——

——

——

——

——

——

——

——

——

0.07

0—

——

——

——

—

Mdh

-2a

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

960

1.00

00.

970

1.00

01.

000

0.53

00.

490

0.70

00.

580

0.84

01.

000

1.00

00.

050

1.00

01.

000

b—

——

——

——

——

——

——

——

——

——

0.04

0—

0.03

0—

—0.

470

0.51

00.

300

0.42

00.

160

——

0.95

0—

—

c—

——

——

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—

d—

——

——

——

——

——

——

——

——

——

——

——

—0.

010

——

——

——

——

—

Pep

-2a

1.00

00.

990

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

b—

0.01

0—

——

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—

Pgi

-1a

0.77

00.

820

0.96

00.

990

1.00

00.

140

1.00

01.

000

—1.

000

1.00

00.

600

0.90

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

990

1.00

00.

920

1.00

00.

990

0.97

00.

850

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

b—

—0.

040

0.01

0—

——

——

——

—0.

100

——

——

——

0.01

0—

——

0.01

00.

030

0.15

0—

——

——

——

—

c0.

230

0.18

0—

——

0.86

0—

—1.

000

——

——

——

——

——

——

0.08

0—

——

——

——

——

——

—

d—

——

——

——

——

——

0.40

0—

——

——

——

——

——

——

——

——

——

——

—


TA

BL

E1.

—C

onti

nu

ed.

A.

mon

tanu

sA

.tr

uei

Nor

ther

nR

ocky

Mts

.S

alm

onR

Mts

.7

Dev

ilsW

allo

wa

Mts

.B

lue

Mts

.N

orth

ern

Cas

cad

esM

ts.

Cen

tral

Cas

cad

esM

ts.

Coa

stM

ts.

Oly

mp

icM

ts.

Sou

ther

nC

asca

des

Mts

.S

iski

you

Mts

.

Pgm

-2a

1.00

01.

000

0.98

00.

850

1.00

01.

000

0.96

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

——

——

—0.

010

——

—0.

010

0.01

0—

——

——

—0.

020

——

b—

—0.

020

0.09

0—

—0.

040

——

——

——

—1.

000

1.00

01.

000

1.00

01.

000

0.99

00.

970

1.00

01.

000

0.99

00.

990

1.00

01.

000

1.00

01.

000

1.00

00.

070

0.98

01.

000

1.00

0

c—

——

0.06

0—

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—

d—

——

——

——

——

——

——

——

——

——

——

——

——

——

—0.

900

0.90

0—

——

e—

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—0.

020

——

—

f—

——

——

——

——

——

——

——

——

——

—0.

030

——

——

——

——

—0.

000

——

—

Pm

i-2

a—

0.92

00.

750

0.70

00.

840

1.00

00.

240

0.04

01.

000

0.36

0—

0.95

0—

——

0.01

0—

0.05

00.

070

0.01

0—

0.01

0—

0.03

00.

180

—0.

200

0.21

00.

030

——

0.02

00.

090

—

b0.

010

—0.

150

0.21

00.

160

—0.

760

0.70

0—

——

0.02

0—

—1.

000

0.99

01.

000

0.95

00.

930

0.99

01.

000

0.98

01.

000

0.88

00.

640

0.92

00.

700

0.76

00.

980

1.00

01.

000

0.98

00.

830

1.00

0

c0.

830

0.08

00.

040

0.09

0—

——

——

——

—1.

000

1.00

0—

——

——

——

——

——

——

——

——

——

—

d—

—0.

020

——

——

0.26

0—

——

——

——

——

——

——

0.01

0—

0.03

00.

060

——

——

——

—0.

090

—

e0.

010

—0.

040

——

——

——

——

——

——

——

——

——

——

——

——

——

——

——

—

f—

——

——

——

——

0.64

01.

000

0.03

0—

——

——

——

——

——

0.03

00.

030

—0.

100

——

——

——

—

g—

——

——

——

——

——

——

——

——

——

——

——

0.02

00.

090

0.08

0—

0.03

0—

——

——

—

Sod

a0.

670

0.71

00.

770

0.74

00.

530

0.75

00.

690

0.53

0—

——

——

——

——

——

——

——

——

——

——

——

——

—

b0.

330

0.29

00.

230

0.26

00.

460

0.25

0—

—1.

000

——

1.00

01.

000

1.00

0—

——

——

——

——

——

——

——

—0.

080

——

—

c—

——

——

—0.

310

0.47

0—

1.00

01.

000

——

—1.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

930

0.02

01.

000

1.00

0

d—

——

—0.

010

——

——

——

——

——

——

——

——

——

——

——

——

——

0.98

0—

—

mea

n

hete

rozy

gosi

ty

0.01

50.

010

0.01

30.

012

0.01

60.

014

0.01

50.

020

0.00

00.

016

0.00

00.

010

0.00

00.

000

0.00

10.

001

0.00

00.

011

0.01

00.

008

0.01

40.

008

0.01

70.

010

0.01

50.

014

0.02

70.

019

0.01

70.

016

0.01

00.

010

0.00

50.

029


combined the new haplotypes discovered inour sampling with the corresponding mtDNAsequences from the 16 A. montanus haplo-types characterized by Nielson et al. (2001)and we included nine representative A. trueisequences. Parsimony analyses provided ini-tial topologies for subsequent evaluation ofML models and estimation of model param-eters (Sullivan and Swofford, 1997; Swoffordet al., 1996). We arbitrarily chose one of tenoptimal trees saved from parsimony searches(equal weights, heuristic search, 10 randomstepwise addition replicates, TBR branchswapping), and held this topology constant inmodel selection using DT-ModSel (Minin etal., 2003) to select from among 56 substitution

models. Each model represents a special caseof the GTR+I+C model of nucleotide evolu-tion, and these are the same models evaluatedby Modeltest (Posada and Crandall, 1998).Abdo et al. (2005) have demonstrated that thisapproach selects simpler models than likeli-hood ratio tests or the AIC, that neverthelessestimate phylogenies accurately. This ap-proach led to the selection of HKY+C forthe 530 bp data set, and ML searches wereperformed with the model fully defined(heuristic search with 10 replicate randomaddition sequences to generate stepwiseaddition trees & TBR branch swapping). MLbootstrap analysis (100 replicates with a max-imum of one tree in each replicate; Felsen-

FIG. 1.—Collection localities for Ascaphus montanus and Ascaphus truei (range shown within inset map; modifiedfrom Metter and Pauken, 1969 and Green and Campbell, 1984). Numbers correspond with the collection localities givenin Appendix I. Sites with a dot indicate mtDNA data. Solid dots indicate localities with newly collected mtDNA, opendots indicate localities with mtDNA sequences reported in Nielson et al. (2001). Open triangles indicate localities whereallozyme data were collected.


stein, 1985) was used to estimate nodalsupport for the phylogeny. In addition, weused a Bayesian approach to estimate poste-rior probabilities for each node. We useduniform priors on all parameters and ran 4chains of 4 3 106 generations each (MRBayes;Huelsenbeck and Ronquist, 2001). We dis-carded the first 10,000 generations as burn-inafter determining that the chains had reachedstable plateaus.

Analysis of Congruence

We used a Bayesian approach to examinecongruence among the allozyme and mtDNAdata. We analyzed the mtDNA data matrixusing MRBayes (Huelsenbeck and Ronquist,2001) under the previously selected HKY+Cmodel of sequence evolution. A run of 6.7 3106 generations was started with a randomtree and we discarded the first 10,000generations as burn-in. We then sampledevery 1000 generations to generate theposterior probability distribution. We filteredthe resulting sample of 5,700 trees to find theportion consistent with the constraint thathaplotypes from the coast (A. truei) and thosefrom the northern Rocky Mountains (A.montanus) exhibit reciprocal monophyly (in-dicated in both allozyme topologies).

We also used the program STRUCTURE(Pritchard, Stephens and Donnelly, 2000) toimplement a non-hierarchal test of the hy-pothesis that A. montanus and A. trueirepresent distinct species. STRUCTURE ap-plies a Bayesian, model-based clusteringmethod to probabilistically assign individualgenotypes to one of K populations. Popula-tions are assumed to be in Hardy-Weinbergequilibrium and linkage equilibrium. A Mar-kov chain Monte Carlo (MCMC) approach isused to estimate model parameters.

We ran STRUCTURE using the ‘‘NoAdmixture Model’’ of individual ancestry. Thismodel assumes that each sampled individual isdrawn discretely from one population, andcalculates the posterior probability that a givenindividual is from each of K populations. Wechose this model based on the lack of long-distance dispersal documented during mark-recapture studies of the tailed frog (Adamsand Frissell, 2001; Daugherty and Sheldon,1982a). For similar reasons we used a model

that assumes independent allele frequenciesamong populations. All runs were 106 genera-tions in length, preceded by a burn-in periodof 10,000 generations (during which parame-ters were verified to have reached stationar-ity). To test the hypothesis that A. montanusand A. truei represent discrete species weexamined the assignment of individual geno-types with K set equal to two.

Phylogeographic Hypothesis Tests WithinA. montanus

We tested three predictions of the westernrefugia hypothesis. We used a two-tailedstudent’s t-test to test the prediction thatpopulations from the Blue and WallowaMountains show a higher mean allozymeheterozygosity. A Bayesian analysis was usedto test the prediction that haplotypes from theBlue and Wallowa Mountains group at theorigin of the A. montanus phylogeny. Wefollowed the methods described above for theanalysis of congruence (i.e., 6.7 3 1026

generations, 10,000 generations discarded asburn-in). We filtered the resulting sample of5700 trees to find the portion consistent withthe constraint that haplotypes from west of theSnake River are basal to the remaining RockyMountains haplotypes. The proportion oftrees in the sample consistent with thehypothesis represents the probability that thehypothesis is correct, given the data, modeland priors.

We used nested clade analysis to examinethe mtDNA data for signs of a range expan-sion from west of the Snake River (Temple-ton, 1998; Templeton et al., 1995). Weconstructed our haplotype network andgrouped clades using guidelines presented inTempleton et al. (1987) and Templeton andSing (1993). We used ParsProb (Posada et al.,2000) to determine the number of substitu-tional steps connecting haplotypes parsimoni-ously. GeoDis (Posada et al., 2000) was usedto calculate the statistics used in nested cladeanalysis (DC, DN, and I-T) and to test theobserved statistics against a null distribution.One thousand permutations of the haplotypenetwork provided a null distribution, repre-senting a random association of haplotypeswith geography, against which to test theempirical values of DC, DN, and I-T for


significance. We use the updated inferencekey (Templeton, 2004) developed in responseto recent critiques of the method (e.g.Knowles and Maddison, 2002) to interpretthe nested clade analysis in terms of phylo-geographic processes.

We used three approaches to test predic-tions of the dual refugia hypothesis. To test forthe association of haplotypes with the pur-ported refugia (and subsequent range expan-sion), we used the nested clade analysisdescribed above. To examine the allozymedata set for structure consistent with twohistoric refugia, we ran STRUCTURE usingthe A. montanus data set for values of K fromone to five. The number of discrete popula-tions (K) can be approximated by running themodel for a range of values, calculating theprobability for each K (from Pr (X|K), andselecting the smallest K from those of similarprobability (because the probability of K oftenplateaus). Three runs were completed foreach value of K. The number of generationsappears to be adequate for approximating Kbecause only slight differences in the value ofln Pr (X|K) were seen within a given K. Modelsettings, the number of generations and theburn-in period are the same as those de-scribed above for the analysis of congruence.

We used MDIV to test whether the date oflineage divergence falls within the Pleisto-cene. This program uses a coalescent-based,divergence by isolation with migration model(Nielsen and Wakeley, 2001). Haplotypesamples were partitioned into two geograph-ically defined populations. We posited thatsamples collected north of the arid SalmonRiver canyon were derived from a Pleistocenerefuge in the Clearwater River valley. Samplescollected south of this break were assumed tohave arisen from a refuge in the South Fork ofthe Salmon River. Samples collected from theSeven Devils, Blue, and Wallowa Mountainsand Craig Mountain were omitted becausetheir predicted origin is unclear with respectthe purported refugia (i.e., they representgeographic isolates of the main range).

The model applied by MDIV uses anMCMC methodology to approximate theposterior distribution of three parameters;the divergence time between populations ingenerations (T), the migration rate in number

of migrants per generation (M), and theta (h;a function of the effective population size andthe nucleotide substitution rate). We useduniform priors and experimented with trun-cation following the recommendations ofNielsen and Wakeley (2001). We completedtwo runs of 2 3 107 generations to assessstationarity. We assumed a nucleotide sub-stitution rate of 1027 (based on the value usedfor amphibian taxa in Carstens et al., 2005)and a generation time of six years (Daughertyand Sheldon, 1982a) to calculate the diver-gence time between the geographically de-limited populations. We summed the frequen-cies of the posterior probabilities in thedistribution for each parameter to define theboundaries that contain 95% of the estimates.The 95% credibility intervals for T served totest the hypothesis that the populations di-verged due to Pleistocene isolation. Estimatesof Ne and M are discussed in the context ofconservation.

RESULTS

Allozymes

Description of loci.—We scored 23 allo-zyme loci in 34 populations; of these, 14 werepolymorphic (Table 1). Heterozygote defi-ciencies were detected at five loci in 11populations (Pgi-1, pops. 14 and 37; Gus,pop. 41; Pmi-2, pops. 42, 35, 47, 15, and 17;Idh-2, pops. 43 and 44; Aat-2, pop. 18).

Several geographically clustered groups ofpopulations exhibit differentiation from theremaining populations at one or more loci.For example, the Cascades Mountains popu-lations are fixed for the ‘‘b’’ allele at the Idh-1locus, whereas other populations carry thisallele at lower frequencies (Table 1). Inaddition, populations from the OlympicMountains are fixed for the ‘‘b’’ allele at theLap locus; this allele does not occur in any ofthe samples from other populations (Table 1).We also found a fixed difference (withnegligible leakage) between A. montanus andA. truei populations at the Pgm-2 locus(Table 1). The A. montanus sample (n 51120 alleles) contains 12 ‘‘A. truei alleles’’(frequency 5 1.07%), whereas the A. trueisample (n 5 1600 alleles) contains 4 ‘‘A.montanus alleles’’ (frequency 5 0.25%).


Phylogeny estimates.—Phylogenetic analy-sis of the allozyme data yields two topologies;Figure 2a corresponds with the analysis thatexcluded the Gus locus, Figure 2b corre-sponds with the analysis that included theGus locus, but excluded population 42 fromthe Salmon River Mts. Both topologiescontain two, well-supported clades corre-sponding to Pacific and Rocky Mountainpopulations (83% and 71% bootstrap values,respectively; Fig. 2). The relationships amongpopulations within these major clades differsomewhat between topology 2a and 2b.Considering only A. truei, in topology 2a,population 56 is sister to all populations foundto the north; however in topology 2b popula-tion 56 is sister only to the group formed bypopulations from the northern CascadesMountains, and the group formed by thecentral Cascades Mountains populations andpopulation 58 from the Coast Mountains. Twogroupings are well supported in both topolo-gies: the group formed by populations 15 and57 from the Siskiyou Mountains, and thegroup formed by populations 18, 50, 51 and52 from the Olympic Mountains (Fig. 2).

Likewise, the relationships among A. mon-tanus populations differ slightly depending onwhether locus Gus is included and population42 is excluded or vice versa. In the analysisexcluding population 42 (but retaining theGus locus; Fig. 2b), the two populations fromthe Wallowa Mountains (pops. 39 and 40)form a clade sister to the rest of A. montanus.Populations from the Blue Mountains (pops.14 and 38) form a clade sister to the remainingRocky Mountains populations. Thus, A. mon-tanus populations from west of the SnakeRiver are paraphyletic to populations east ofthe river. However, inclusion of population 42and the elimination of the Gus locus alters thisconclusion (Fig. 2a). Well supported clades inboth allozyme topologies include: populations39 and 40 from the Wallowa Mountains,populations 14 and 38 from the Blue Moun-tains, and populations 33 and 34 from thePioneer and Seven Devils Mountains (Fig. 2).

Population level parameters.—We removedthree loci (Gus, Aat-2 and Ldh-2) from allAMOVA runs because Arlequin cannot per-form an AMOVA with loci resolved in fewerthan 95% of the individuals genotyped. Fifty-

five percent of the allozyme variation parti-tions between A. montanus and A. trueipopulations, (FCT 5 0.55; P , 0.001), 28%occurs among populations within groups (FSC

5 0.63; P , 0.001), and 17% occurs withinpopulations (FST 5 0.83; P , 0.001).

Out of 572 comparisons, 20 pairwise FST

values are not significant (P . 0.05). Weaklydifferentiated A. truei populations include thethree northernmost populations in the Cas-cades Mountains (pops. 23, 48 and 59), threepopulations from the western Olympic Moun-tains (pops. 50, 51 and 52), and populationsfrom the Coast Mountains in Oregon and thecentral Cascades Mountains (pops. 17, 44, 45,48, 49, 53, 56, and 58; Fig. 3). Weaklydifferentiated A. montanus populations in-clude two from the Blue Mountains (pops. 14and 38), two from the Seven-Devils Moun-tains (pops. 39 and 40), and two from theBitterroot Mountains (pops. 41 and 36;Fig. 3).

mtDNA Within A. montanus

Our sampling identifies seven additional A.montanus haplotypes distinct from the 16previously characterized (Nielson et al., 2001).With this additional sampling, we definea southern clade within A. montanus distin-guished by up to 0.031 substitutions per sitefrom northern populations (Fig. 4).

All parsimony trees measure 90 steps (70variable sites; 52 parsimony-informative; con-sistency index [CI] 5 0.833; retention index[RI] 5 0.959; rescaled consistency index [RC]5 0.799). The HKY+C model, which allowsunequal base frequencies, and in which alltransitions are assigned a different rate ofsubstitution from all transversions, is themodel chosen by decision theory (Minin etal., 2003). Parameter estimates for the MLtree under the HKY+C model are as follows:transition to transversion rate ratio 5 4.2;shape parameter for the C2distribution 51.6; base frequencies pA 5 0.272; pC 5 0.243;pG 5 0.162; pT 5 0.323.

All optimal parsimony topologies, and theML topology contain two major haplotypeclades. The northern clade is comprised ofhaplotypes from populations north of theSouth Fork of the Salmon River and popula-tions in the Blue, Wallowa, and Seven Devils


Mountains. The southern clade is composedof populations south of the South ForkSalmon River in the Boise and Salmon RiverMountains. Alternative parsimony topologiesdiffer from each other and the ML topology

only in the resolution of internal brancheswithin each clade (in particular haplotype FFis alternatively placed sister to the rest of thenorthern clade; Fig. 4). Bootstrap values andBayesian posterior probabilities strongly sup-port monophyly of haplotypes from southernIdaho (98% and 1.00, respectively). Theremaining A. montanus haplotypes group withlow support because of the alternate position-ing of haplotype FF (12% bootstrap, 0.41posterior probability; Fig. 4). Maximum-likeli-hood corrected genetic distances (HKY+C)between these groups average 0.024 substitu-tions per site (range 0.018–0.031). The meancorrected genetic distance within the southernhaplotype group is 0.004 (range 0.002–0.006);mean distance within the northern clade is0.006 (range 0.002–0.012).

Analysis of Congruence

Each of the 5700 trees in the Bayesianposterior distribution exhibits reciprocalmonophyly between Rocky Mountains andPacific populations. Thus, the two charactersystems (mtDNA and allozymes) are strongly

FIG. 2.—(A) Maximum-likelihood tree estimated from the allozyme frequency data using CONTML with locus Gusomitted and rooted at the midpoint. (B) Maximum-likelihood tree estimated from the allozyme frequency data usingCONTML with the Dollar Cr. population (42) omitted. Population numbers (given in parentheses) and geographicdesignations correspond with Appendix I and Fig. 1. Neighbor-joining bootstrap values (200 replicates) based onCavalli-Sforza chord distances (Cavalli-Sforza and Edwards, 1967) are shown above branches for those clades withgreater than 50 percent support.

FIG. 3.—Map showing populations with nonsignificantallozyme FST values. Dashed lines connect populationswithout significant FST values. Population numberscorrespond with Appendix A and Fig. 1. A table of FST

values is given in Appendix III.


concordant (P 5 1.0) with respect to re-ciprocal monophyly across the disjunction.Comparison of the mtDNA phylogeny to theallozyme topologies is not straight-forwardbecause the tips of the allozyme topologyrepresent populations and those of themtDNA phylogeny represent unique haplo-types. Nevertheless, haplotypes from the samelocale or population as those represented inthe allozyme-based phylogeny do not group inthe same manner (Figs. 2 and 4).

When the number of populations (K) is setequal to two, each of the 1085 allozymegenotypes is assigned with high probability toeither the A. montanus or the A. truei group.Out of 471 individuals sampled from theRocky Mountains, 455 group with A. mon-tanus with a probability of one; the remaining16 genotypes are assigned to A. montanus withprobabilities ranging from 0.999 to 0.998.Thirteen individuals assigned with probabili-ties , 1.0 originate from populations 32 and

42; one originates from each of populations36, 39, and 40. Out of 614 individuals sampledfrom the Coast and Cascades Mountains, 607are assigned to A. truei with a probability ofone. The remaining seven (all from population55) are assigned with probabilities rangingfrom 0.869 to 0.999.


We examined three predictions under thewestern refugia hypothesis. The first, thatmtDNA haplotypes from the Blue and Wal-lowa Mountains will group at the origin of theA. montanus phylogeny, is not supported byour analysis. Only twelve of 5700 treesgenerated for the Bayesian analysis areconsistent with colonization of the northernRocky Mountains from populations west ofthe Snake River (P 5 0.002).

The second prediction, that mean allozymeheterozygosity will be higher within popula-tions from the Blue and Wallowa Mountains,is not supported by our data either. Meanheterozygosity is actually lower within popula-tions from the Blue and Wallowa Mountainswhen compared with populations from theremainder of the species’ range (P 5 0.05).

The third prediction, that mtDNA halo-types will show evidence of a range expansionfrom west of the Snake River is not validatedby the nested clade analysis. The nested cladenetwork includes haplotypes connected by upto eight substitutional steps (Fig. 5). Haplo-types K, A and EE are found in populationsfrom the Blue and Wallowa Mountains. Thesehaplotypes group within the northern clade(clade 2-2). With respect to the northerngroup, haplotypes K and EE appear geo-graphically restricted (P , 0.001). HaplotypeA, which is interior, and thus ancestral, in thenorthern group’s haplotype network, appearswidespread geographically within the north-ern portion of A. montanus’ range (P , 0.001;Fig. 5). The group of haplotypes closelyrelated to haplotype A (clade 1-1), showsa pattern consistent with restricted gene flowwith isolation by distance. Based on theupdated nested-clade inference key (Temple-ton, 2004), the restricted ranges of haplotypesK and EE suggest one or more past fragmen-tation events.

FIG. 4.—One of five A. montanus maximum-likelihoodtrees estimated from the cyt b data set under theobjectively determined HKY+C model of sequenceevolution and rooted using nine representative A. trueisequences. This tree has a lnL of 21273.85 and differsfrom the other maximum-likelihood and parsimony treesonly in alternative resolutions of internal branches.Numbers above branches represent maximum-likelihoodbootstrap values (100 replicates). Those below branchesrepresent Bayesian estimates of nodal support (4 chains of107 generations each). The populations where eachhaplotype was found are indicated in parentheses.Population numbers and geographic designations corre-spond with Fig. 1 and Appendix I.


We examined four predictions under thedual refugia hypothesis. The first, that A.montanus will contain two monophyleticclades, is consistent with the mtDNA topology(Fig. 4). The allozyme topology, however,does not show two clades within A. montanus(Fig. 2). When the number of populations (K)is set to two, individuals are not clearlyassigned to northern or southern clades.Individuals from the four northernmost po-pulations (5, 36, 37, 41, and 43) are assigned ina mixed manner to both groups without highprobabilities. Individuals from three popula-tions in the Blue and Wallowa Mountains (14,38, and 35) are assigned exclusively to groupone, while individuals from populations 39 and40 are assigned exclusively to group two. Theremaining individuals from populations in thecentral and southern portion of the speciesrange are assigned in a clear manner to eithergroup one or two; however, this assignmentshows no geographic pattern. Values of K fromone to five do not produce an optimal lnPr(X|K). This indicates the presence of morethan five populations in the sampled geno-types. This result is not unexpected given thenumber of significant pairwise FST values forA. montanus (Appendix III).

The second and third predictions, that themtDNA haplotypes will show a geographicassociation consistent with their purported

refugia and that the clades will show evidenceof range expansion, were examined usingnested clade analysis. Haplotypes are dividedinto two major nesting clades that correspondwith the northern and southern clades iden-tified in the ML topology (Fig. 6). Randomassociation of haplotypes with geographycould be rejected for the northern group(clade 2-2; P , 0.001). The dispersion ofhaplotypes did not appear strongly geograph-ically structured in the southern group (clade2-1; P 5 0.073). With respect to the northerngroup, haplotypes D, C, K and FF and the E,F, G group appear geographically restricted (P5 0.008, P 5 0.006, P , 0.001, P , 0.001, andP , 0.001, respectively). The group formed byhaplotypes one-step removed from haplotypeA (clade 1-1) shows a restricted geographicrange clustered near the center of thenorthern group (P , 0.001; Fig. 5).

Using the updated inference key (Temple-ton, 2004), four clades contained sufficientsampling and genetic structure to infer pro-cesses. The group of haplotypes closely re-lated to haplotype A (clade 1-1), and thoseclosely related to haplotype O (clade 1-8)show patterns consistent with restricted geneflow with isolation by distance. Within thenorthern clade (clade 2-2) the restrictedranges of several geographically peripheralhaplotypes (FF, E, F, G, K, EE) indicate oneor more past fragmentation events. When thenorthern and southern clades (clade 2-2 and2-1, respectively) are considered together thegeographic pattern is consistent with pastfragmentation followed by range expansion.Thus, the nested clade analysis supports thesecond and third predictions of the dualrefugia hypothesis.

The final prediction, that the divergencebetween northern and southern populationswill date to the Pleistocene, was examinedusing MDIV. Our estimate of the timing oflineage divergence, 999,902 years ago (95%credibility interval 5.4 mya to 92,892 yearsago), supports a Pleistocene split. This isconsistent with the dual refugia hypothesis.

DISCUSSION

The allozyme data we present providesfurther support for the recognition of A.montanus as originally proposed by Nielson

FIG. 5.—The minimum-spanning network of haplotypesgrouped into nesting clades following procedures given inTempleton et al. (1987) and Templeton and Sing (1993).Each dash represents a substitution. Zeros indicateancestral haplotypes that were not sampled. Connectionsup to 9 steps have a 95% probability of a parsimoniousconnection. White cells represent one-step clades andshading encloses two-step clades. Clade numbers aregiven within each enclosure with the level of the nestingclade followed by the number within that level (i.e. 2-1,nesting level two, grouping number one). Haplotypeletters correspond with those given in Figs. 4 and 6.


et al. (2001) based on monophyly within themtDNA lineage. According to the unifiedgeneral lineage concept of species (deQueiroz, 1998), the strongest support fora given species hypothesis will rise frommultiple species criteria. Monophyly withinthe mtDNA phylogeny represents one line ofevidence under the general lineage concept ofspecies. The monophyly criterion is also metby the allozyme phylogeny. Fixation of thePgm-2 locus within A. montanus meets thediagnosability criterion (again using terminol-ogy introduced by de Queiroz, 1998). Thecorrect assignment of each of the 1085allozyme genotypes to either A. montanus orA. truei with high probability also supportsthis criterion. Furthermore, the majority(55%) of allozyme variation within tailed frogspartitions between A. montanus and A. truei

as indicated by the AMOVA. Congruencebetween the allozyme and mtDNA phyloge-nies (with respect to the interspecies di-vergence) suggests concordance between themultiple gene geneaolgies approximated bythe allozyme phylogeny and the mtDNA genelineage. These independent phylogenies pro-vide evidence that both mtDNA and nucleargenes coalesce more recently within A.montanus than between A. montanus and A.truei, and thus satisfy the concordant co-alescence criterion.

Additional differences in the nuclear ge-nome were found by Ritland et al. (2000) incomparisons of A. truei and A. montanus inBritish Columbia using randomly amplifiedpolymorphic DNA (RAPD); the deepest di-vergence in RAPDs also occurred betweenRocky Mountain and Pacific samples. Al-though limitations imposed by a lack ofunderstanding of the mechanisms behindRAPD evolution prevent direct comparisonswith our mtDNA and allozyme data, thesedifferences between Pacific and Rocky Moun-tain populations in British Columbia areconsistent with recognition of A. montanusas a distinct species.

Allozyme Variation Within A. truei

Both allozyme topologies show three well-supported geographic groupings within A.truei: the clade formed by two populationsfrom the Siskiyou Mountains of Oregon(populations 15 and 57), the clade formed bypopulations 23, 46, and 59 from the northernCascades Mountains, and the clade formed bypopulations from the Olympic Mountains(populations 18, 50, 51, and 52; Fig. 2). FST

values indicate gene flow among populationswithin the northern Cascades Mountainsclade and the Olympic Mountains clade, andindicate a lack of genetic exchange amongeach of these three clades and the remainingpopulations from the central Cascades Moun-tains and Coast Mountains of Oregon (Fig. 3).

Populations from the Coast Mountains andcentral Cascades Mountains of Oregon andWashington (populations 17, 44, 45, 47, 48,49, 53, 56, and 58) do not group distinctly bygeography in the allozyme topology. FST

values indicate a lack of divergence inallozyme frequencies among these popula-

FIG. 6.—The dispersion of A. montanus haplotypes ongeography. Haplotypes are indicated at their collectionlocality and correspond with the designations given inFig. 4. Collection localities correspond with Fig. 1 andAppendix I. The northern and southern clades areenclosed by shading, and the range of the RockyMountain tailed frog is indicated in the inset map.


tions. Nielson et al. (2001) did not samplewithin this central part of the species’ rangeand could not determine whether mtDNApatterns were consistent with early Pleisto-cene refugia or limited dispersal and isolationby distance. The presence of non-significantFST values and lack of geographic groupingwithin the core of the Pacific tailed frog’srange suggests that relatively recent dispersalmay have occurred and/or that contemporarygene flow has not been limited.

Unlike many populations to the north,pairwise FST values among populations fromthe Siskiyou Mountains and Southern Cas-cades Mountains in Oregon show significantdifferentiation of allele frequencies (Fig. 3;Appendix III). Significant FST values mayreflect reduced gene flow due to climate-induced limitations on dispersal. Tailed frogsexhibit greater mobility over land duringprecipitation events and their habitats becomeincreasingly fragmented in dry climates(Daugherty and Sheldon, 1982b). Similaritiesin precipitation and climate within the south-ern portion of the Pacific tailed frog’s rangeand the range of the Rocky Mountain tailedfrog may explain why FST values within thesesouthernmost populations exhibit a patternsimilar to that seen within A. montanus, wheresignificant differentiation is found amongnearly all population pairs. An alternative(although not exclusive) explanation for thispattern may be that a range expansionoccurred earlier within the southern portionof the Pacific tailed frog’s range as comparedto the northern part, such that allozymefrequencies have had additional time todiverge among southern populations. Thisexplanation is consistent with the directionof glacial retreat. Under this hypothesis,contemporary isolation could be similarthroughout all parts of the species’ range.


Nielson et al. (2001) postulated that A.montanus populations persisted within thenorthern Rocky Mountains through the Pleis-tocene. Carstens et al. (2005) estimated themean time of divergence between the twoAscaphus lineages at 3.1 3 106 years ago,further supporting this hypothesis. Pliocene

drought and subsequent Pleistocene glacialcycles provide a context for describing con-temporary genetic patterns within A. monta-nus. During the Pliocene the northern RockyMountains began to experience summerdrought and lower mean annual temperatures(Wolfe, 1969, 1978). Pleistocene glacial cyclesfollowed these climate changes. The Cordil-leran ice-sheet periodically covered the RockyMountain region north of Coeur d’Alene,Idaho, and montane glaciers extended souththroughout the Rocky Mountain tailed frog’srange during glacial maxima from 2 million to12,000 years ago (Alt and Hyndman, 1995).

The western refugia hypothesis asserts thathabitats west of the Snake River served asa refuge for A. montanus through thePleistocene while the body of their currentrange in the northern Rocky Mountains wasglaciated. This hypothesis is not supported byour data. The placement of populations fromthe Blue and Wallowa Mountains at the originof A. montanus within allozyme topology 2bappears consistent with the western refugiahypothesis; however, this placement is notwell supported, and is not stable to inclusion/omisision of the Gus locus. Furthermore, ourBayesian analysis indicated that haplotypesfrom the Blue and Wallowa Mountains do notconsistently group basally in the mtDNAphylogeny. Mean allozyme heterozygosity islower within populations west of the SnakeRiver compared to the remainder of thespecies’ range, rather than higher as wouldbe expected for populations with a longerhistory in the area (given that population sizeand interpopulation genetic exchange appearto be similar). Nested clade analysis indicatesthat unique haplotypes from the Blue andWallowa Mountains arose as a result ofhistoric fragmentation and isolation; signalsof range expansion from west of the SnakeRiver are not evident in the data.

The Clearwater River drainage has beensuggested as a potential refuge for mesicforest species during the Pleistocene becauseof the large numbers of endemic and coastaldisjunct plant species found here (Dauben-mire, 1975). Explicit hypothesis testing basedon mtDNA sequence data within the Coeurd’Alene salamander (Plethodon idahoensis)supports expansion of this species from the


Clearwater drainage at the close of thePleistocene (Carstens et al., 2004a). Similaranalysis of sequences from the Idaho giantsalamander (Dicamptodon atterimus) indi-cates expansion from the South Fork of theSalmon River (Carstens et al., 2004b). Boththe Coeur d’Alene salamander and the Idahogiant salamander have ecological tolerancessimilar to those of the tailed frog. The dualrefugia hypothesis posits that isolated popula-tions of the Rocky Mountain tailed frogpersisted within the Clearwater and SalmonRiver Valleys during glacial maxima. Much ofthe existing data support this hypothesis.

The dual refugia hypothesis predicts thatallozyme and mtDNA data will be partitionedinto two groups, that these groups will showa geographic affinity for the purported re-fugia, that signals of a range expansion fromthe refugia will be evident in the data, and thatthe timing of divergence between the cladeswill correspond with a Pleistocene division ofthe lineage. The A. montnaus mtDNA phy-logeny is composed of two monophyleticclades, distinguished by 1.8–3.1 percent se-quence divergence, with a strong north/southgeographic association. Nested clade analysisindicates that these groups arose due tohistoric fragmentation and subsequent rangeexpansion. Post-glacial expansion is reflectedin the dispersion of haplotypes within eachclade. Both clades are represented by onecommon haplotype: A in the northern cladeand O in the southern clade. These haplotypesare the most frequently occurring in eachclade and are found in almost every localitysampled. Haplotypes A and O are ancestral tomultiple derived haplotypes in the phylogeny,and the derived haplotypes are found to a largeextent along the periphery of the sampledrange (Figs. 5, 6). Because little contemporarygene flow occurs among populations withineach clade (based on the significant differen-tiation indicated by allozyme FST analyses;Fig. 3; Appendix II) the distribution of eachcommon haplotype suggests a range expansionconsistent with colonization during favorablepost-glacial periods.

The timing of lineage divergence betweenthe northern and southern clades is consistentwith glacially induced isolation. Our pointestimate of the timing of lineage divergence

(999,902 years ago) falls in the mid-Pleisto-cene. The upper bound of the credibilityinterval (92,829 years ago) remains within thePleistocene. The lower bound (5.4 million -years ago) extends into the late Pliocene. Thedual refugia hypothesis emphasizes the rolethat glacial cycles and cooling trends played tofoster geographic isolation. Based on ourcredibility interval we cannot exclude therole that increasing aridity through thePliocene may have also played in isolatingthese groups.

In contrast, the allozyme phylogeny doesnot show two distinct clades. Only populations32 and 42 are from localities within the rangeof the southern mtDNA clade. These popula-tions do not group together in the allozymetopologies, nor do they carry unique allelesdistinguishing them from the remaining A.montanus populations (Fig. 2; Table 1). Fur-thermore, a nonhierarchical analysis of theallozyme data showed that individual geno-types could not be clearly assigned to twogroups with a north/south geographic affinity.Our data show no indication that a southernallozyme clade would emerge given moreextensive sampling south of the Salmon RiverMountains; populations 32 and 42 carry themost common allele at 13 of the 14 loci.Secondary contact and/or male mediated geneflow could explain the absence of distinctallozyme clades. However FSTvalues amongpopulations 32 and 42 and all remainingpopulations indicate a lack of current geneflow (Fig. 3; Fig. 6).

Conservation

Conservation within A. truei.—The Pacifictailed frog is listed as a species of specialconcern by state and provincial governmentsin California, Oregon, Washington and BritishColumbia. Although local distribution de-pends on a number of variables, Ascaphuspopulations are sensitive to increased siltationand elevated water temperatures that mayaccompany land management activities suchas timber harvest and road building (Corn andBury, 1989; Walls et al., 1992; Welsh, 1990).This has generated concern over the loss andfragmentation of old growth habitat in thePacific Northwest and the effect this may haveon populations of tailed frogs (Blaustein et al.,


1994; Bury, 1983; Corn and Bury, 1989; Wallset al., 1992; Welsh, 1990).

The concept of evolutionarily significantunits (ESUs) is designed to aid long-termconservation planning at an intraspecific levelby delineating population groups that repre-sent significant elements of genetic diversityin the lineage (Avise, 1989; Ryder, 1986). TheESU concept is valuable and practical for thePacific tailed frog because much of its habitatis publicly administered by agencies witha mandate to conserve biological diversity(and limited resources to do so). Furthermore,substantial genetic diversity is present withinthe lineage. Pairwise FST values show a mini-mum of seven discrete groups within A. truei,and data from Nielson et. al. (2001) show fourmtDNA clades distinguished by one to threepercent sequence divergence.

Several defining criteria have been pro-posed to delimit ESUs, these include: uniqueecological adaptations (Dizon et al., 1992;Waples, 1991), diagnosable characters (Voglerand DeSalle, 1994), geographic isolation, anda distinct phylogenetic lineage (Vogler andDeSalle, 1994). These criteria may overlapwith those used to delimit species under somephylogenetic species concepts (Cracraft,1991). We choose to describe ESUs usingthe following criteria introduced by Moritz(1994): ESUs should be reciprocally mono-phyletic for mtDNA alleles and show signifi-cant divergence of allele frequencies at nuclearloci. Under these criteria nuclear and mito-chondrial evidence is required to preventmisclassifying populations linked by male-mediated gene flow and, because of the largenumber of generations necessary to acquirereciprocal monophyly, only divergence ofallele frequencies is required for nuclear loci.

The lack of uniform geographic samplingfor mtDNA creates an obstacle to definingESUs within the Pacific tailed frog; howeverwe present a preliminary working delineationof these groups pending further sampling.This species appears to be represented bya minimum of two ESUs: (1) OlympicMountains populations and (2) populationsfrom the Umpqua River south through theSiskiyou Mountains and into northern Cali-fornia. The Olympic Mountains ESU exhibitsa unique fixed allele at the Idh-1 locus and

is represented by a monophyletic clade inthe mtDNA phylogeny (Nielson et al., 2001).The Siskiyou Mountains/southern CascadesMountains clade is supported by significantFST values, divergence of frequencies at theAat-2 and Idh-1 loci, and a monophyleticmtDNA clade (Nielson et al., 2001). FST

values indicate significant divergence of allo-zyme frequencies between these two ESUsand the remaining populations in the Coastand Cascades Mountains; however it is notclear whether the remaining populations alsoform a single ESU. Haplotypes from thenorthern Cascades Mountains and CoastMountains do not form a clade in the mtDNAphylogeny, and mtDNA sequences are notavailable for populations from the centralCascades (Nielson et al., 2001).

In addition to ESUs, pairwise FST valuesmay communicate useful conservation infor-mation because they provide an indicator ofgenetic connectivity among populations. Non-significant FST values indicate contemporaryor recent historical association among popula-tions. Based on pairwise FST values the threenorthernmost populations (23, 46, and 59) andthe Olympic Mountains populations (18, 50,51, and 52) of the Pacific tailed frog areisolated from the remaining populations in thecentral Cascades Mountains and Coast Moun-tains. Recent historical or contemporarygenetic exchange appears to connect popula-tions from the Coast and central CascadesMountains. In contrast, each populationsampled south of the Umpqua River appearsisolated. As described earlier, this pattern maybe due to the effects of an increasingly aridclimate on dispersal in the southern latitudes.

Conservation within A. montanus.—TheRocky Mountain tailed frog is listed as a statespecies of concern in Washington and isfederally protected as endangered in BritishColumbia. The species has yet to garnerconservation priority in Idaho or Montana.The range of A. montanus is characterized bymultiple mountain ranges separated by in-creasingly xeric valleys toward its southernextent. Presumably due to geographic andclimatic isolation, pairwise FST values showsignificant genetic differentiation for all butthe nearest neighboring populations (Fig. 3;Appendix III). This genetic isolation, in


combination with minimal dispersal documen-ted in mark recapture studies (Adams andFrissell, 2001; Daugherty and Sheldon,1982b) emphasizes the need to managepopulations on a local watershed or subwa-tershed scale.

Under Moritz’s (1994) criteria the RockyMountain tailed frog is represented by twoESUs: (1) the northern clade that includespopulations from the Blue, Wallowa and SevenDevils Mountains and the northern RockyMountains from the Sesech River north, and(2) the southern clade that encompasses thespecies range south of the South Fork of theSalmon River. Each of these groups formsa monophyletic clade in the mtDNA phylog-eny. FST values (which reflect allele frequen-cies at multiple nuclear loci) indicate signifi-cant divergence of allozyme frequenciesbetween northern and southern populations.The recognition of two ESUs within A.montanus could be questioned because almostall population pairs show significant FST valuesand the allozyme phylogeny does not containdistinct northern and southern clades. Never-theless, we choose to delimit these groupsbecause each contains a unique and significantportion of the Rocky Mountain tailed frog’smtDNA diversity, and our data indicate a lackof genetic exchange between the groups.Furthermore, allozyme sampling includedonly two populations within the geographicboundaries of the southern ESU, which mayhave confounded our analysis.

We were able to estimate the effectivepopulation size for the Rocky Mountain tailedfrog and the number of migrants exchangedbetween the northern and southern cladesusing a coalescent-based, divergence by iso-lation with migration model implemented byMDIV (Nielsen and Wakeley, 2001). Based onthis model (and assuming a mutation rate of1 3 1027, used in Carstens et al., 2005), theRocky Mountain tailed frog is represented byan effective population size of 51,436 females(95 percent credibility interval extending from22,118 to 92,584). The census number ofadults would be much higher than thisestimate of effective population size forseveral reasons. First, a maximum of half ofthe female population breeds each year, suchthat the number of females in the population

is at least twice the effective population size(Daugherty and Sheldon, 1982a). Further-more, the model assumes low variance inreproductive success, a stable population size,and no assortative mating; violations of theseassumptions could certainly result in a censuspopulation an order of magnitude greater thanthe effective population size. We suspect thatreproductive success varies widely amongyears and between streams in response tothe magnitude and timing of flow events(Lohman, 2002; Metter, 1968), but strongdocumentation for this is lacking. Similarly,there have likely been fluctuations in popula-tion size in response to changes in climaticconditions and habitat availability. Unfortu-nately, evaluating our estimate of effectivepopulation size in relation to an estimate ofthe actual number of tailed frogs in thenorthern Rockies is hampered, as it is formost amphibians, by the general scarcity ofpopulation data. A few studies have focusedon defining the range of A. montanus ina given area (e.g., Dupuis et al., 2000;Marnell, 1997). A number of studies haveestimated larval densities for tailed frogs(Dupuis and Stevenson, 1999; Hawkins et al,1988; Lohman, 2002), but estimates of adultnumbers have been rare. Daugherty andSheldon (1982a) marked 543 individuals inan 80 m section of Butler Creek (Montana)and Metter (1964) reported capture rates ashigh as 0.4–1.1 individuals/m in sections ofstreams in the Touchet River (Washington)and Palouse River (Idaho) drainages. Al-though these estimates do suggest that tailedfrogs can be locally abundant, an estimate ofpopulation size will require a much morecomprehensive understanding of densitiesacross the range.

Based on MDIV results 10 migrants areexchanged between the northern and south-ern clades each year (95% credibility interval6–35). This estimate appears reasonable basedon the high degree of philopatry that has beenobserved in the Rocky Mountain tailed frog(Adams and Frissell, 2001; Daugherty andSheldon, 1982b) and the relatively fragmentednature of mesic forests along the zone ofcontact between these clades. This estimate ofexchange is consistent with the significantpairwise FST values found in this species.


Based on minimal recaptures of newly meta-morphosed frogs (Daugherty and Sheldon,1982b), some uncertainty has existed regard-ing long-distance dispersal in juvenile tailedfrogs. Both genetic markers confirm that long-range dispersal either does not occur withinthis juvenile cohort or does not contributesignificantly to genetic exchange.

Acknowledgments.—Special thanks go to C. and A.Berklund for financial support through the University ofIdaho College of Natural Resources and the Berklundresearch assistantship. We thank R. Glesne (NorthCascades National Park), B. Hossack (Glacier NationalPark) and M. Adams (Olympic National Park) forproviding tissue samples from their respective nationalparks. D. Posada and K. Crandall provided valuableinformation on the use of GeoDis software. J. Felsensteinprovided information on the use of PHYLIP software. N.Nielson assisted with figure preparation. Facilities used inthis research were supported by grant DHHS/NIHRR11833 to the Department of Biological Sciences atthe University of Idaho and by the NSF EPSCoR programand NSF cooperative agreement number EPS-9720634(to JS). Additional support was provided by a University ofIdaho Research Office seed grant to K. Lohman.Collection and processing of samples for mtDNA wasdone under animal care and use protocol 9005 andWashington scientific collection permit number 99-189.Tissues from mtDNA collection are deposited in theUniversity of Idaho vertebrate collection. DNA sequencesare deposited in the GenBank database under acces-sion numbers AF 277324–AF277352 and DQ087511–DQ087517.

LITERATURE CITED

ABDO, Z., V. MININ, P. JOYCE, AND J. SULLIVAN. 2005.Accounting for uncertainty in the tree topology has littleeffect on the decision theoretic approach to modelselection in phylogeny estimation. Molecular Biologyand Evolution 22:691–703.

ADAMS, S. B., AND C. A. FRISSELL. 2001. Thermal habitatuse and evidence of seasonal migration by RockyMountain tailed frogs, Ascaphus montanus, in Mon-tana. The Canadian Field Naturalist 115:251–256.

ALT, D., AND D. W. HYNDMAN. 1995. Northwest exposures:a geologic story of the Northwest. Mountain PressPublishing Company, Missoula, Montana, U.S.A.

AVISE, J. C. 1989. A role for molecular genetics in therecognition and conservation of endangered species.Trends in Ecology and Evolution 4:279–281.

BLAUSTEIN, A. R., D. B. WAKE, AND W. P. SOUSA. 1994.Amphibian declines: judging stability, persistence, andsusceptibility of populations to local and global extinc-tions. Conservation Biology 8:60–71.

BRUNSFELD, S. J., J. SULLIVAN, D. E. SOLTIS, AND P. S.SOLTIS. 2001. Comparative phylogeography of north-western North America: A Synthesis. Pp. 319–339. InIntegrating ecology and evolution in a spatial context.J. Silvertown and J. Antonovics (Eds.). BlackwellPublishing, Williston, Vermont, U.S.A.

BURY, R. B. 1983. Differences in amphibian populations inlogged and old growth redwood forest. NorthwestScience 57:167–178.

CANNATELLA, D. C., AND D. M. HILLIS. 1993. Amphibianrelationships: Phylogenetic analysis of morphology andmolecules. Herpetological Monographs 7:1–7.

CARSTENS, B. C., A. L. STEVENSON, J. D. DEGENHARDT, AND

J. SULLIVAN. 2004a. Testing nested phylogenetic andphylogeographic hypotheses in the Plethodon vandykeispecies group. Systematic Biology 53:781–792.

———. 2004b. Accounting for coalescent stochasticity intesting phylogeographical hypotheses: modeling Pleis-tocene population structure in the Idaho giant sala-mander Dicamptodon aterrimus. Molecular Ecology14:255–265.

CARSTENS, B. C., S. J. BRUNSFELD, J. R. DEMBOSKI, J. M.GOOD, AND J. SULLIVAN. 2005. Investigating the evolu-tionary history of the Pacific Northwest mesic forestecosystem: hypothesis testing within a comparativephylogeographic framework. Evolution, In Press.

CAVALLI-SFORZA, L. L., AND A. EDWARDS. 1967. Phyloge-netic analysis: models and estimation procedures.Evolution 21:550–570.

CLAYTON, J., AND D. TRETIAK. 1972. Aminocitrate buffersfor pH control in starch gel electrophoresis. Journal ofthe Fish Research Board of Canada 29:1169–1172.

COCKERHAM, C. C. 1969. Variance of gene frequencies.Evolution 23:72–84.

———. 1973. Analyses of gene frequencies. Genetics74:679–700.

CORN, P. S., AND R. B. BURY. 1989. Logging in westernOregon: responses of headwater habitats and streamamphibians. Forest Ecology Management 29:39–57.

CRACRAFT, J. 1991. Systematics, species concepts, andconservation biology. Abstract. Fifth Annual Meeting ofthe Society of Conservation Biology. University ofWisconsin, Madison, Wisconsin, U.S.A.

DAUGHERTY, C. H. 1979. Population ecology and geneticsof Ascaphus truei: an examination of gene flow andnatural selection. Ph.D. Dissertation, University ofMontana, Missoula, Montana, U.S.A.

DAUGHERTY, C. H., AND A. L. SHELDON. 1982a. Age-determination, growth, and life-history of a Montanapopulation of the tailed frog (Ascaphus truei). Herpe-tologica 38:461–468.

———. 1982b. Age-specific movement patterns of adultAscaphus truei. Herpetologica 38:468–474.

DAUBENMIRE, R. 1975. Floristic plant geography of easternWashington and northern Idaho. Journal of Biogeogra-phy 2:1–18.

DE QUEIROZ, K. 1998. The general lineage concept ofspecies, species criteria, and the process ofspeciation: Aconceptual unification and terminological recommen-dations. Pp. 57–75. In Endless Forms: Species andSpeciation. D. J. Howard and S. H. Berlocher (Eds.).Oxford University Press, Oxford, England.

DIZON, A. E., C. LOCKYER, W. F. PERRIN, D. P. DEMASTER,AND J. SISSON. 1992. Rethinking the stock concept.Conservation Biology 6:24–36.

DUPUIS, L. A., AND D. STEVENSON. 1999. Riparianmanagement and the tailed frog in northern coastalforests. Forest Ecology and Management 124:35–43.


DUPUIS, L. A., F. L. BUNNELL, AND P. A. FRIELE. 2000.Determinants of the tailed frog’s range in BritishColumbia, Canada. Northwest Science 74:109–115.

FELSENSTEIN, J. 1981. Evolutionary trees from DNAsequences: a maximum likelihood approach. Journalof Molecular Evolution 17:368–376.

———. 1985. Confidence limits on phylogeny: anapproach using the bootstrap. Evolution 39:783–791.

FORD, L. S., AND D. C. CANNATELLA. 1993. The majorclades of frogs. Herpetological Monographs 7:94–117.

FROST, D. R. 2004. Amphibian Species of the World: anOnline Reference. Version 3.0 (22 August 2004).Electronic Database accessible at http://research.amnh.org/herpetology/amphibia/index.php. AmericanMuseum of Natural History, New York, USA.

GREEN, D. M., AND R. W. CAMPBELL. 1984. Theamphibians of British Columbia. British ColumbiaProvincial Museum, Victoria, British Columbia, Ca-nada.

GUO, S., AND E. THOMPSON. 1992. Performing the exacttest of Hardy-Weinberg proportion for multiple alleles.Biometrics 48:361–372.

HAWKINS, C. P., L. J. GOTTSCHALK, AND S. S. BROWN. 1988.Densities and habitat of tailed frog tadpoles in smallstreams near Mt. St. Helens following the 1980eruption. Journal of the North American BenthologicalSociety 7:246–252.

HAY, J. M., I. RUVINSKY, S. B. HEDGES, AND L. R. MAXSON.1995. Phylogenetic relationships of amphibian familiesinferred from DNA sequences of mitochondrial 12Sand 16S ribosomal RNA genes. Molecular Biology andEvolution 12:928–937.

HUELSENBECK, J. P., AND F. RONQUIST. 2001. MRBAYES:Bayesian inference of phylogenetic trees. Bioinfor-matics 17:754–755.

KNOWLES, L. L., AND W. P. MADDISON. 2002. Statisticalphylogeography. Molecular Ecology 11:2623–2635.

LOHMAN, K. 2002. Annual variation in the density ofstream tadpoles in a northern Idaho (USA) watershed.Verhandlungen der Internationalen Vereinigung furTheoretische und Angewandte Limnologie 28:802–806.

MARNELL, L. F. 1997. Herpetofauna of Glacier NationalPark. Northwestern Naturalist 78:17–33.

METTER, D. E. 1964. A morphological and ecologicalcomparison of two populations of the tailed frog,Ascaphus truei Stejneger. Copeia 1964:181–195.

———. 1968. The influence of floods on populationstructure of Ascaphus truei Stejneger. Journal ofHerpetology 1:105–106.

METTER, D. E., AND R. J. PAUKEN. 1969. An analysis of thereduction of gene flow in Ascaphus truei in theNorthwest U.S. since the Pleistocene. Copeia1969:307–310.

MININ, V., Z. ABDO, P. JOYCE, AND J. SULLIVAN. 2003.Performance-based selection of likelihood models forphylogeny estimation. Systematic Biology 52:674–683.

MORITZ, C. 1994. Defining evolutionarily significant unitsfor conservation. Trends in Ecology and Evolution9:373–375.

NIELSEN, R., AND J. W. WAKELEY. 2001. DistinguishingMigration from Isolation: an MCMC Approach. Ge-netics 158:885–896.

NIELSON, M. K., L. K. LOHMAN, AND J. SULLIVAN. 2001.Phylogeography of the tailed frog (Ascaphus truei):implications for the biogeography of the Pacific North-west. Evolution 55:157–160.

PAUKEN, R. J., AND D. E. METTER. 1971. Geographicrepresentation of morphological variation among popu-lations of Ascaphus truei Stejneger. Systematic Zoology20:431–441.

POSADA, D., AND K. A. CRANDALL. 1998. Modeltest: testingthe model of DNA substitution. Bioinformatics14:917–918.

POSADA, D., K. A. CRANDALL, AND A. R. TEMPLETON. 2000.GeoDis: a program for the cladistic nested analysis ofthe geographical distribution of genetic haplotypes.Molecular Ecology 9:487–488.

RIDGWAY, G., S. SHERBURNE, AND R. LEWIS. 1970. Poly-morphism in the esterases of the Atlantic herring.Transactions of the American Fisheries Society99:147–151.

RITLAND, K., L. A. DUPUIS, F. L. BUNNELL, W. L. HUNG,AND J. E. CARLSON. 2000. Phylogeography of the tailedfrog (Ascaphus truei) in British Columbia. CanadianJournal of Zoology 78:1749–1758.

RYDER, O. A. 1986. Species conservation and systematics:the dilemma of subspecies. Trends in Ecology andEvolution 1:9–10.

SCHNEIDER, S., D. ROESSLI, AND L. EXCOFIER. 2000.Arlequin: a software for population genetics dataanalysis. Vers. 2.000. Genetics and Biometry Lab,Department of Anthropology, University of Geneva,Geneva, Switzerland.

SICILIANO, M., AND C. SHAW. 1976. Separation andvisualization of enzymes on gels. In I. Smith (Ed.),Chromatographic and Electrophoretic Techniques,Vol. 2, Zone Electrophoresis. Heinemann, London,U. K.

SULLIVAN, J., AND D. L. SWOFFORD. 1997. Are guinea pigsrodents? The importance of adequate models inmolecular phylogenetics. Journal of Mammalian Evo-lution 4:77–86.

SWOFFORD, D. L. 1998. PAUP*: phylogenetic analysisusing parsimony (*and other methods). Vers. 4. SinauerAssociates, Sunderland, Massachusetts, U.S.A.

SWOFFORD, D. L., G. J. OLSEN, P. J. WADDELL, AND D. M.HILLIS. 1996. Phylogenetic inference. Pp. 407–514. InD. M. Hillis, C. Moritz, and B. K. Mable (Eds.),Molecular systematics, Second edition. Sinauer Associ-ates, Sunderland, Massachusetts, U.S.A.

TEMPLETON, A. R. 1998. Nested clade analyses ofphylogeographic data: testing hypotheses about geneflow and population history. Molecular Ecology7:381–397.

———. 2004. Statistical phylogeography: methods ofevaluating and minimizing inference errors. MolecularEcology 13:789–810.

TEMPLETON, A. R., AND C. F. SING. 1993. A cladisticanalysis of phenotypic associations with haplotypesinferred from restriction endonuclease mapping. IV.Nested analyses with cladogram uncertainty and re-combination. Genetics 134:659–669.

TEMPLETON, A. R., E. BOERWINKLE, AND C. F. SING. 1987.A cladistic analysis of phenotypic associations withhaplotypes inferred from restriction endonucleasemapping. I. Basic theory and an analysis of alcohol


dehydrogenase activity in Drosophila. Genetics117:343–351.

TEMPLETON, A. R., E. ROUTMAN, AND C. A. PHILLIPS. 1995.Separating population structure from population histo-ry: a cladistic analysis of the geographical distributionof mitochondrial DNA haplotypes in the tiger salaman-der, Ambystoma tigrinum. Genetics 140:767–782.

VOGLER, A. P., AND R. DESALLE. 1994. Diagnosing units ofconservation management. Conservation Biology8:354–363.

WALLS, S. C., A. R. BLAUSTEIN, AND J. J. BEATTY. 1992.Amphibian biodiversity of the Pacific Northwest withspecial reference to old growth stands. The NorthwestEnvironmental Journal 8:53–69.

WAPLES, R. S. 1991. Pacific salmon, Oncorhynchus spp.,and the definition of ‘‘species’’ under the EndangeredSpecies Act. Marine Fisheries Review 53:11–22.

WELSH, H. H. 1990. Relictual amphibians and old growthforests. Conservation Biology 4:309–319.

WOLFE, J. A. 1969. Neogene floristic and vegetationalhistory of the Pacific Northwest. Madrono 20:83–110.

———. 1978. A paleobotanical interpretation of Tertiaryclimates in the Northern Hemisphere. AmericanScientist 66:694–703.

.Accepted: 24 May 2006

.Associate Editor: Kevin de Quieroz

APPENDIX I

Collection localities. Population numbers correspond to those from Figures 1, 2 and 4, and where applicable, tocollection localities given in Nielson et al. (2001). Populations 1a and 1b and 12a and 12b were considered singlepopulations in Nielson et al. (2001). Populations are marked in column D for mtDNA data and column A for allozyme

data.

Population # Data County Mountain Range Watershed Stream

D A1a X Idaho. Latah Co. Palouse

MountainsPalouse R. Mountain Gulch

1b X Idaho. Latah Co. PalouseMountains

Palouse R. Mannering Cr.

2 X Idaho. Idaho Co. Clearwater Mts. Lochsa R. Indian Post Office Cr.3 X Idaho. Clearwater Co. Clearwater Mts. Little N.

Fk. ClearwaterUnnamed Tributary

4 X Idaho. Idaho Co. Clearwater Mts. S. Fk Clearwater Legett Cr.5 X X Montana. Missoula Co. Bitterroot Mts. Clark Fk. R. East Fork Lolo Cr.6 X Idaho. Nez Perce Co. Craig Mt. Snake R. Eagle Cr.7 X Montana. Flathead Co. Livingston Mts. Middle Fk.

Flathead R.Lower Rubridge Cr.

8 X Montana. Flathead Co. Livingston Mts. Middle Fk.Flathead R.

Autumn Cr.

9 X Montana. Flathead Co. Livingston Mts. Middle Fk.Flathead R.

Upper Fern Cr.

10 X Montana. Glacier Co. Livingston Mts. St. Mary R. Reynolds Cr.11 X Idaho. Valley Co. Salmon River

Mts.Sesech R. Maverick Cr.

12a X Idaho. Valley Co. Salmon RiverMts.

East Fk. of theS. Fk. Salmon R.

Parks Cr.

12b X Idaho. Valley Co. Salmon RiverMts.

East Fk. of theS. Fk. Salmon R.

Reegan Cr.

13 X Idaho. Idaho Co. Salmon RiverMts.

Salmon R. Slate Cr.

14 X X Washington.Columbia Co.

Blue Mts. Touchet R. Headwaters Touchet R.

15 X X Oregon. Josephine Co. Siskiyou Mts. Illinois R. Sucker Cr.16 X Oregon. Josephine Co. Siskiyou Mts. Illinois R. Lost Canyon Cr.17 X X Oregon. Benton Co. Oregon Coast Mts. Alsea R. Parker Cr.


APPENDIX IContinued.

Population # Data County Mountain Range Watershed Stream

18 X X Washington. Mason Co. Olympic Mts. Hamma Hamma R. Jefferson Cr.19 X Washington. Clallam Co. Olympic Mts. Drains to

Pacific OceanEnnis Cr.

20 X Washington. Skagit Co. Cascade Mts. Skagit R. Happy Cr.21 X Washington. Chelan Co. Cascade Mts. Stehekin R. Bridge Cr.22 X Washington. Skagit Co. Cascade Mts. Stehekin R. McAllister Cr.23 X X Washington. Chelan Co. Cascade Mts. Wenatchee R. Smith Br.24 X Idaho. Valley Co. Salmon River Mts. E. Fk. of the

S. Fk. Salmon R.Moose Cr.

25 X Idaho. Valley Co. Salmon River Mts. E. Fk. of the S.Fk. Salmon R.

Sheep Cr.

26 X Idaho. Valley Co. Salmon River Mts. Middle Fk.Payette R.

Unnamed headwatertributary

27 X Idaho. Valley Co. Salmon River Mts. N. Fk. Payette R. Clear Cr.28 X Idaho. Valley Co. Salmon River Mts. S. Fk. of the

Salmon R.Curtis Cr.

29 X Idaho. Custer Co. Sawtooth Range Yankee Fk. ofthe Salmon R.

Five Mile Cr.

30 X Idaho. Valley Co. Salmon River Mts. S. Fk. of theSalmon R.

Four Mile Cr.

31 X Idaho. Valley Co. Salmon River Mts. S. Fk. of theSalmon R.

Buckhorn Cr.

32 X X Idaho. Elmore Co. Boise Mts. Middle Fk. Boise R. Unnamed tributarynear Rocky Bar

33 X X Montana. Beaverhead Co. Pioneer Mts. Big Hole R. Steele Cr.34 X X Idaho. Washington Co. Seven Devils Mts. Snake R. Brownlee Cr.35 X X Oregon. Wallowa Co. Wallowa Mts. Imnaha R. Lick Cr.36 X Montana. Mineral Co. Bitterroot Mts. Clark Fk. R. Oregon Falls Cr.37 X Montana. Glacier Co. Livingston Mts. Flathead R. Sprague Cr.38 X Washington. Garfield Co. Blue Mts. Snake R. Tucannon R.39 X Oregon. Baker Co. Wallowa Mts. Powder R. O’Brien Cr.40 X Oregon. Wallowa Co. Wallowa Mts. Grand Ronde R. Lostine R.41 X Idaho. Shoshone Co. Clearwater Mts. St. Joe R. Bird Cr.42 X Idaho. Valley Co. Salmon River Mts. S. Fk. Salmon Dollar Cr.43 X Montana. Missoula Co. Bitterroot Mts. Clark Fk. R. Butler Cr.44 X Washington. Skamania Co. Cascade Mts. White Salmon R. Mosquito Cr.45 X Washington. Kittitas. Co. Cascade Mts. Keechelus Lk. Hyak Cr.46 X Washington. Skagit Co. Cascade Mts. S. Fk. Sauk R. Weden Cr.47 X Washington. Cowlitz Co. Cascade Mts. Toutle R. Coldwater Cr.48 X Washington. Kittitas Co. Cascade Mts. S. Fk. Snoqualmie R. Surveyor Cr.49 X Washington. Cowlitz Co. Cascade Mts. Toutle R. Elk Cr.50 X Washington. Grays

Harbor Co.Olympic Mts. Humptulips R. W. Fk. Humptulips R.

51 X Washington.Grays Harbor Co.

Olympic Mts. Quinault R. Fletcher Canyon

52 X Washington.Grays Harbor Co.

Olympic Mts. Quinault R. Merriman Cr.

53 X Oregon. Linn Co. Cascade Mts. Lost Lk. Hackleman Cr.54 X Oregon. Douglas Co. Cascade Mts. Umpqua R. Bear Cr.55 X Oregon. Douglas/Jackson

Co. lineCascade Mts. Rogue R. Bert Cr.

56 X Oregon. Lane Co. Cascade Mts. S. Fk. McKenzie R. Starr Cr.57 X California. Humboldt Co. Coast Mts. Redwood Cr.58 X Oregon. Tillamook Co. Coast Mts. Nestucca R. Mt. Hebo59 X Washington. Glacier Co. Cascades Mts. Nooksack R. Boyd Cr.


APPENDIX II

Summary of electrophoretic conditions of the allozyme loci included in this study. The following abbreviations are usedfor the tissues: L 5 liver, M 5 muscle, T 5 tadpole.

Enzyme System Locus DesignationEnzyme Commission

Number Buffer System1 Tissue Alleles identified

Asparatate aminotransferase Aat-1Aat-2

2.6.1.1 A M, T Aat-1 (a, b)A M, T Aat-2 (a, b, c, d)

a-Glycerophosphatedehydrogenase

Agp-1 1.1.1.8 C M, T Agp-1 (a, b)

B-glucuronidase Gus 3.2.2.31 B L, T Gus (a, b, c)Isocitrate dehydrogenase Idh-1 1.1.1.42 C M, T Idh-1 (a, b)

Idh-2 B L, T Idh-2 (a, b)Leucine aminopeptidase Lap 3.4.11.1 A M, T Lap (a, b)Lactate dehydrogenase Ldh-2 1.1.1.27 C L, T Ldh-2 (a, b)Malate dehydrogenase Mdh-2 1.1.1.37 B L, T Mdh-2 (a, b, c, d)Peptidase Pep-2 3.4.–.– A L, T Pep-2 (a, b)Phosphoglucose isomerase Pgi-1 5.3.1.9 A L, T Pgi-1 (a, b, c, d)Phosphoglucomutase Pgm-2 5.4.2.2 A M, T Pgm-2 (a, b, c, d, e, f)Phosphomannose isomerase Pmi-2 5.3.1.8 A M, T Pmi-2 (a, b, c, d, e, f, g)Superoxide dismutase Sod 1.15.1.1 A L, T Sod (a, b, c, d)

1 Buffer A (Ridgway et al., 1970) gel: 0.03 M Tris, 0.005 M citric acid, pH 8.5; electrode: 0.06 M lithium hydroxide, 0.3 M boric acid, pH 8.1; gels were madeusing 99% gel buffer and 1% electrode buffer; Buffer B (Clayton and Tretiak, 1972) gel: 0.002 M citric acid, pH 6.0; electrode: 0.04 M citric acid, pH 6.1; bothbuffers are pH adjusted with N-(3-Aminopropyl)-Morpholine; Buffer C (Siciliano and Shaw, 1976) gel: 0.009 M Tris, 0.003 M citric acid, pH 7.0. Gels weremade using 14% starch in the appropriate gel buffer.


AP

PE

ND

IXII

IP

airw

ise

FS

Tva

lues

for

all

pop

ula

tion

ssa

mp

led

for

allo

zym

es.

Pop

ula

tion

nu

mb

ers

corr

esp

ond

wit

hth

ose

inF

ig.

1an

dA

pp

end

ixI.

Sig

nif

ican

tF

ST

valu

essh

own

inb

old

.

514

1517

1823

3233

3435

3637

3839

4041

4243

4445

4647

4849

5051

5253

5455

5657

58

5

14

0.6

1

15

0.7

80.9

4

17

0.6

90.8

40.5

9

18

0.7

60.8

70.6

30.6

4

23

0.8

40.9

80.9

30.4

70.8

2

32

0.3

50.7

50.8

0.6

20.6

90.8

4

33

0.4

40.8

60.9

10.8

10.8

40.9

50.6

4

34

0.6

70.9

50.9

70.8

90.9

10.8

40.7

2

35

0.4

0.8

40.9

50.8

60.8

90.9

80.7

50.7

30.7

9

36

0.0

60.6

80.8

30.7

10.7

80.8

80.2

80.4

70.7

90.6

1

37

0.3

70.4

40.8

40.7

30.8

0.8

70.4

50.5

50.7

70.6

50.2

8

38

0.5

70.

060.9

50.8

30.8

61

0.7

30.8

81

0.9

0.6

60.4

3

39

0.5

40.9

0.9

0.7

0.7

80.9

80.5

70.8

10.9

80.9

10.5

90.6

20.9

5

40

0.5

70.9

20.9

10.7

10.7

81

0.6

0.8

31

0.9

40.6

20.6

31

0.07

41

0.0

60.6

20.7

80.6

60.7

60.8

30.2

50.4

40.7

20.5

30

0.2

90.5

90.5

40.5

6

42

0.3

70.7

0.7

40.5

80.7

20.8

0.0

70.6

30.8

10.7

20.3

20.4

40.6

70.4

70.4

90.3

43

0.0

60.7

10.8

50.7

50.8

10.8

90.4

30.3

50.7

10.5

20.0

70.3

60.7

0.6

50.6

80.0

80.4

6

44

0.8

10.9

40.8

50.4

0.7

90.0

70.8

0.9

10.9

60.9

40.8

50.8

50.9

40.8

90.9

0.8

10.7

70.8

6

45

0.6

70.9

40.7

92

0.15

0.6

50.1

70.6

50.8

70.9

90.9

50.7

20.7

40.9

80.8

60.9

20.6

50.5

80.7

70.

03

46

0.8

30.9

70.9

20.4

60.8

20

0.8

40.9

51

0.9

80.8

80.8

71

0.9

70.9

90.8

30.8

0.8

90.0

60.0

9


514

1517

1823

3233

3435

3637

3839

4041

4243

4445

4647

4849

5051

5253

5455

5657

58

470.8

10.9

50.8

80.4

10.7

90.0

70.8

0.9

20.9

70.9

50.8

50.8

50.9

60.9

10.9

30.8

0.7

70.8

60.0

60.7

70.0

6

480.7

40.9

50.8

60.3

50.7

40.2

80.7

50.8

90.9

80.9

60.7

90.8

0.9

70.8

90.9

10.7

40.7

0.8

20.1

20.

020.2

30.1

1

490.7

60.9

50.8

50.3

40.7

40.1

70.7

60.9

0.9

80.9

60.8

0.8

10.9

70.9

0.9

30.7

50.7

10.8

32

0.01

0.0

40.1

40.0

80.1

500.7

40.9

30.7

10.6

80.1

40.9

40.7

20.8

60.9

70.9

50.7

70.8

0.9

30.8

10.7

90.7

40.7

20.8

10.8

60.7

0.9

30.8

80.7

80.8

1

510.7

20.8

80.6

90.5

80.1

10.8

50.6

30.8

20.9

20.9

0.7

30.7

70.8

70.7

50.7

40.7

10.6

80.7

80.7

90.5

80.8

40.7

90.7

0.7

20.1

7

520.7

0.8

10.5

60.5

60.0

60.7

40.6

10.7

80.8

40.8

20.7

10.7

40.7

90.6

90.6

90.6

90.6

60.7

40.7

20.5

50.7

40.7

10.6

60.6

62

0.12

0.01

530.7

10.8

70.6

30.0

40.6

30.6

40.6

40.8

40.9

20.8

90.7

30.7

50.8

70.7

50.7

60.6

80.6

0.7

70.5

50.

060.6

30.5

60.5

0.5

0.7

0.5

80.5

4

540.7

0.8

60.7

10.4

90.7

10.7

10.6

50.8

20.9

0.8

70.7

20.7

50.8

40.7

20.7

30.6

90.6

0.7

60.6

40.4

90.7

0.6

50.5

80.5

70.7

20.6

80.6

50.5

4

550.6

90.8

80.5

40.5

30.6

50.8

20.6

60.8

50.9

20.8

90.7

30.7

60.8

70.7

70.7

80.6

80.5

80.7

70.7

40.6

30.8

20.7

70.7

20.7

0.6

90.6

70.6

0.5

70.2

3

560.6

90.9

0.7

12

0.01

0.6

60.5

90.6

50.8

50.9

50.9

20.7

30.7

50.9

10.7

90.8

0.6

70.5

90.7

80.4

22

0.15

0.5

80.4

40.3

80.3

80.7

20.6

0.5

60.0

60.5

10.5

9

570.7

20.9

50.2

60.5

40.5

50.9

70.7

40.8

90.9

90.9

60.7

70.7

90.9

80.8

90.9

20.7

20.6

70.8

10.8

40.7

60.9

60.8

80.8

20.8

40.5

20.5

70.5

0.5

80.6

50.5

0.6

5

580.7

20.9

40.8

10.2

50.6

80.4

80.7

20.8

80.9

90.9

60.7

60.7

80.9

70.8

70.9

0.7

10.6

60.8

10.

022

0.06

0.4

10.1

80.1

40

0.6

80.6

30.6

0.4

10.5

10.6

30.2

60.7

3

590.8

40.9

80.9

30.4

70.8

20

0.8

40.9

51

0.9

80.8

80.8

71

0.9

81

0.8

30.8

0.8

90.0

70.1

70

0.0

70.2

80.1

70.9

40.8

50.7

40.6

40.7

10.8

20.5

90.9

70.4

8

AP

PE

ND

IXII

IC

onti

nu

ed.


Documents

HERPETOLOGICA - University of Idahojacks/Nielsonetal06.pdf · United States and southwestern Canada. These frogs display a number of morpholog-ical adaptations to their unique environments