american marten ( martes americana ) in the pacific ...msb.unm.edu/isles/small_et_al 2003 pacific nw...

Molecular Ecology (2003)

12

89ndash103

copy 2003 Blackwell Publishing Ltd

Blackwell Science Ltd

American marten (

Martes americana

) in the Pacific Northwest population differentiation across a landscape fragmented in time and space

MAUREEN P SMALL

KAREN D STONE

dagger

and JOSEPH A COOK

Dagger

WDFW Genetics Division 600 Capitol Way N Olympia WA 98501ndash1091 USA

dagger

Department of Biology Southern Oregon University 1250 Siskiyou Blvd Ashland OR 97520 USA

Dagger

Biological Science Idaho State University Pocatello ID 83209 USA

Abstract

American marten (

Martes americana

) have a close association with mature temperateforests a habitat that expanded throughout the Pacific Northwest as glaciers receded at theend of the Pleistocene Similar to several other forest-associated mammals in North America(eg black bear) genetic analysis of the marten shows a deep phylogeographical sub-division that reflects populations with distinctive evolutionary histories Using a suite of14 microsatellite markers we explored the genetic structure of marten populations intwo reciprocally monophyletic clades in the Pacific Northwest identified previously as

M caurina

and

M americana

by mitochondrial haplotypes and morphology Microsatellitephylogeographical patterns were congruent with mitochondrial analyses These independ-ent data sets shed light upon hybridization patterns population structure and evolutionaryhistories Hybridization between

M caurina

and

M americana

individuals was documentedin two regions of sympatry (Kuiu Island in southeastern Alaska and southern Montana)Northern insular populations of

M caurina

exhibited higher differentiation and lowervariability relative to northern populations of

M americana

Greater divergence among

M caurina

populations may reflect longer isolation and persistence in coastal forest habitatthat was fragmented by rising sea level in the early Holocene Lower differentiation amongnorthern

M americana

populations and close relationship to other continental

M americana

populations may reflect more recent expansion into the Pacific Northwest andor continuedgene flow among populations Differentiation among

M caurina

populations was attrib-uted to habitat fragmentation (ie rising sea level) as opposed to isolation-by-distanceoceanic straits pose significant barriers to gene flow among

M caurina

populations andbetween populations of

M caurina

and

M americana

Keywords

hybridization

Martes americana

microsatellites Pacific Northwest phylogeographysoutheastern Alaska

Received 21 May 2002 revision received 26 September 2002 accepted 26 September 2002

Introduction

The retreat of the Wisconsin glaciers beginning about12 000 years ago was followed by dynamic northwardcolonization of boreal forests and associated speciesinto high latitude regions of North America This biome isbecoming a focus of investigations related to the impactof current deforestation and habitat fragmentation onwildlife In the last 100 years this region has experienced

intensive logging that has changed habitat (Strickland ampDouglas 1987) and prey base (Thompson 1994) for forest-associated carnivores such as marten (

Martes americana

)Genetic-based investigations of population structure(Mitton amp Raphael 1990 McGowan

et al

1999 Kyle

et al

2000) have begun to explore dispersal patterns gene flowand genetic diversity in marten A genetic frameworkprovides an opportunity to explore the dynamic historicalbiogeography of this medium-sized carnivore and topredict the organismrsquos response to habitat change thusfurther delimiting units for more effective conservationand management in regions experiencing current

Correspondence Maureen P Small Fax 360 902 2943 E-mailmosmallislandnetcom

90

M P S M A L L K D S T O N E and J A C O O K


Molecular Ecology

12 89ndash103

widespread anthropogenic disturbance (Crozier 1992Ryder 1986 Vane-Wright

et al

1991 Faith 1992 Moritz 1994)Although eight subspecies of

M americana

have beendescribed (Clark

et al

1987) these are traditionally placedinto two morphological groups

caurina

and

americana

(Merriam 1890 Clark

et al

1987) Molecular investigationscorroborate the distinction of

caurina

and

americana

as tworeciprocally monophyletic mitochondrial clades (Carr ampHicks 1997 Stone

et al

2002) These forms appear to havediverged due to isolation in distinct southern glacial refugia(

caurina

populations isolated in western and

americana

populations isolated in eastern United States respectivelyCarr amp Hicks 1997 Stone amp Cook 2002 Stone

et al

2002)because there is little evidence of a high latitude refugiumduring the Wisconsin that was forested (Pielou 1991)Youngman (1993) suggests the persistence of

americana

in Beringia during the Pleistocene but material that heexamined was not dated either stratigraphically or through

isotope analysis and may have been deposited during theHolocene

The two forms are largely allopatric with

caurina

popu-lations documented along the West Coast (California tosoutheastern Alaska) and then eastward to WyomingMontana and Idaho (Fig 1 Wright 1953 Hall 1981 Carr ampHicks 1997 Stone

et al

2002) Currently

americana

popu-lations are more widespread distributed from Montanaand Idaho northward to Alaska and eastward to the Atlan-tic Coast Two zones of contact have been identified oneextending through southern and western Montana and theother on Kuiu Island in southeastern Alaska (Stone

et al

2002 Wright 1953 Stone amp Cook 2002) Although Merriam(1890) describes these groups as distinctive species Wright(1953) reports morphological intergradation betweenthe groups in Montana and suggested that they belongto a single species

M americana

a conclusion reflectedby the current taxonomy (Wilson amp Reeder 1993) Carr amp

Fig 1 Map of sampling locations alsoshowing distribution of mitochondrialclades in American martens (Martesamericana) Light and dark grey shadingrepresents regions inhabited by membersof the americana and caurina clades re-spectively with striped regions indicatingcontact zones Inset map shows the NorthAmerican distribution of martens modifiedfrom Hall (1981)

P O P U L A T I O N S T R U C T U R E I N A M E R I C A N M A R T E N S

91


Molecular Ecology

12 89ndash103

Hicks (1997) however questioned whether gene flow existsbetween these groups and concluded that the

caurina

and

americana

clades represent distinct speciesWe investigated genetic differentiation among marten

from 20 localities throughout the Pacific Northwest ofNorth America using 14 nuclear microsatellite markersWe focus on southeastern Alaska because both

caurina

and

americana

populations occur in this region of highendemism (Cook amp MacDonald 2001) Southeastern Alaskahas long been recognized as a unique biogeographicalregion (Swarth 1936 Klein 1965) with numerous spe-cies and subspecies identified nominally as endemics(MacDonald amp Cook 1996) This region is an importantnatural laboratory for investigating the effects of dispersaland habitat fragmentation during the Holocene informationuseful for predicting the effects of habitat fragmentation onpopulation structure and viability (McCauley 1993) Forexample molecular phylogeographical studies have demon-strated that several mammalian species are representedby distinct evolutionary lineages in southeastern Alaska(Talbot amp Shields 1996 Demboski

et al

1999 Conroy ampCook 2000 Stone amp Cook 2000 Cook

et al

2001 Fleming ampCook 2002 Stone

et al

2002) This common pattern indi-cates that each species colonized the region on multipleoccasions and from multiple source populations follow-ing glacial retreat Stone

et al

(2002) hypothesized that

M caurina

populations represent an early Holocene coloniza-tion northward along the coast as coastal ice receded at theend of the last glaciation whereas

americana

populationsrepresent a later colonization from continental source popu-lations that expanded through river corridors traversingthe coastal mountains Coastal mountains were deglaciatedmuch later than the Alexander Archipelago (Pielou 1991)Several sympatric species show similar patterns of differ-entiation (Cook

et al

2001)This study explores the distinctive histories of

caurina

and

americana

populations using nuclear microsatellitemarkers We investigated differentiation of marten popu-lations in the Pacific Northwest in light of potentiallyephemeral corridors and barriers to colonization of morerecently deglaciated areas We consider the dynamic his-tory of the area and the role of glaciers changing sea levelsand geographical distance in isolating populations or inpromoting contact between these distinctive forms ofAmerican marten

Materials and methods

Sampling

Twenty localities (hereafter referred to as populations 10island 10 mainland Fig 1) represented by 413 individualswere chosen centring on the Pacific Northwest region(40ndash52

deg

N 113ndash126

deg

W) of North America (12 populations

from southeastern Alaska four from British Columbiaone from interior Alaska two from Montana and one fromOregon Fig 1 Table 1) With the exception of northernBritish Columbia (

n

= 5) each population was representedby 11ndash26 individuals (Table 1) Previously described areasof sympatry between

caurina

and

americana

populations(Stone

et al

2002) were represented by southern Montana(

n

= 11) and Kuiu Island southeastern Alaska (

n

= 25) Threeisland populations from southeastern Alaska (ChichagofBaranof and Prince of Wales Islands) were the result ofhuman introductions in the mid-1900s by the Alaska GameCommission (Elkins amp Nelson 1954 Burris amp McKnight1973)

DNA extraction and microsatellite amplification

DNA was extracted from marten tissues (heart spleenskeletal muscle skin or blood) archived in the AlaskaFrozen Tissue Collection of the University of AlaskaMuseum (AFTC) Methods of NaCl extraction followedthose of Fleming amp Cook (2002) All samples had beenscreened previously to determine mtDNA clade profiles(

americana

or

caurina

) using automated sequencing orrestriction fragment length analysis (Stone

et al

2002)Amplifications of microsatellite DNA contained 023 m

m

of each primer 154 m

m

each dNTP 14 25 or 43 m

m

MgCl

2

25 mgmL Bovine Serum Albumin (BSA) 025 unitsof DNA polymerase Perkin-Elmer 1

times

PCR buffer II and50ndash100 ng genomic DNA We used a Perkin-Elmer GeneAmpPCR System 9700 with the following PCR conditions onecycle of 94

deg

C for 1 min (for AmpliTaq) or one cycle of 95

deg

Cfor 12 min (for AmpliTaq Gold) followed by two cycles (30s at 94

deg

C 20 s at 58

deg

C 5 s at 72

deg

C) 33 cycles (15 s at 94

deg

C20 s at 54

deg

C 5 s at 72

deg

C) and one cycle (30 min at 72

deg

C)The following primers were used MA1 MA2 MA3

MA5 MA8 MA11 MA14 MA15 MA18 MA19 (Davis ampStrobeck 1998) MER041 MVIS020 MVIS072 and MVIS075(Fleming

et al

1999) Perkin-Elmer AmpliTaq DNA poly-merase was used with most primers with the exception ofMA3 MA19 MER041 MVIS020 MVIS072 and MVIS075where AmpliTaq Gold DNA polymerase was used Thefinal concentration of MgCl

2

for most reactions was 43 m

m

with the exception of reactions using primers MA8 MA15and MVIS020 (14 m

m

MgCl

2

) and MER041 MVIS072 andMVIS075 primers (25 m

m

MgCl

2

) Samples were run onan ABI 373 automated sequencer and alleles were sized(basepairs bp) using an internal lane size standard (GS350by Perkin-Elmer) GeneScan Analysis 31 and Genotyper11 computer programs

Data analysis

fstat

293 (httpwwwunilchizeasoftwaresfstathtml)(Goudet 2001) was used to calculate descriptive statistics

92



Molecular Ecology

12 89ndash103

for populations including mean number of alleles allelicrichness (number of alleles corrected for sample size)observed and expected heterozygosity and

F

IS

valuesSignificance of population

F

IS

values was tested with28 000 randomizations

genepop

version 33 (httpwwwcefecnrs-mopfr) (Raymond amp Rousset 1995) wasused to test for HardyndashWeinberg equilibrium (HWE) andgenotypic linkage disequilibrium Deviations from HWEand linkage disequilibrium were tested per locus andbetween each pair of loci For loci with four or fewer allelesexact tests (Louis amp Dempster 1987) were used to estimate

P

-values to test for deviations from HWE Loci with greaterthan four alleles had

P

-values estimated by the Markovchain method (Guo amp Thompson 1992) Genotypic linkagedisequilibrium was tested using the Markov chain methodwith 10 000 dememorization 5000 batches and 10 000iterations Results were corrected using a sequentialBonferroni adjustment (initial

α

= 005) for multiplecomparisons

An unrooted network of genetic relationships amongindividuals was inferred from allele-sharing distances(Bowcock

et al

1994) Pairwise distances were calculatedusing

sharedst

(httpwwwbiologyualbertacajbrzustosharedsthtml) with the allele-sharing distance defined asone minus half the average number of shared alleles

per locus The

fitch

program in

phylip

version 35c(httpevolutiongeneticswashingtoneduphyliphtml)(Felsenstein 1993) was used to construct a Fitch amp Margoliash(1967) tree from the allele-sharing distance matrix

Pairwise chord distances (Cavalli-Sforza amp Edwards1967) among populations and population networks weregenerated from allele frequency data using

phylip

Chorddistances were calculated in

gendist

and used to constructa neighbour-joining (NJ) tree in the program

neighbour

To test the robustness of tree topologies 1000 bootstrapreplicates of the allele frequency file were generated in

seqboot

and analysed in

gendist

Tree topologies werecreated for all replicates using

neighbour

and a consensustree was generated in

consense

In addition Neirsquos geneticdistance (Nei 1972) was calculated and used in a NJ ana-lysis A maximum likelihood analysis was also performedusing

seqboot contml

and

consense

to generate abootstrapped maximum likelihood (ML) tree A principal-components analysis (PCA) was performed upon popula-tions using

pcagen

(httpwwwunilchizeasoftwarespcagenhtml) Populations were ordinated according to thefirst second and third PCA axes

A Bayesian clustering method structure (Pritchardet al 2000) was used to examine population structure andassign individuals to inferred population clusters based

Table 1 Descriptive statistics for americana and caurina populations of American martens including population locations and abbreviations(abbr) mtDNA clade profile (mtDNA) sample size (n) mean number of alleles (alleles) mean allelic richness (richness) expectedheterozygosity (HE) observed heterozygosity (HO) FIS (bold out of HWE) and FST for the clade comparison

Populations abbr mtDNA n alleles richness HE HO FIS FST

Baranof Island SE AK BAR americana 26 364 25 039 045 026Northern British Columbia NBC americana 5 386 39 066 066 007central British Columbia CBC americana 17 529 404 068 070 016Chichagof Island SE AK CHIC americana 25 429 32 053 057 016Cleveland Penninsula SE AK CP americana 25 536 365 065 065 012Juneau SE AK JUN americana 25 507 345 058 061 019Kupreanof Island SE AK KUP americana 25 443 31 055 056 015Mitkof Island SE AK MIT americana 25 471 33 052 058 023northern Montana NMT americana 11 456 37 062 065 019Prince of Wales Island SE AK POW americana 25 443 32 055 061 026Revillagigedo Island SE AK REV americana 25 364 28 047 049 024Thomas Bay SE AK TB americana 20 429 33 058 062 018Yakutat SE AK YAK americana 22 40 28 048 051 013Yukon Flats interior AK AK americana 25 557 35 062 062 014Kuiu Island SE AK KUIU caurina 25 486 315 050 052 019southern Montana SMT caurina 11 529 42 059 068 026Admiralty Island SE AK ADM caurina 25 129 122 006 007 008Queen Charlotte Islands BC QCI caurina 11 193 18 024 026 004Vancouver Island BC VAN caurina 23 293 21 032 032 013Oregon OR caurina 17 336 27 048 049 minus002americana clade 301 33 058 047 0196 014caurina clade 112 25 037 031 0144 042

SE AK = southeastern Alaska USA BC = British Columbia CanadaMixed = both M americana and M caurina mitochondrial DNA haplotypes

P O P U L A T I O N S T R U C T U R E I N A M E R I C A N M A R T E N S 93

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 89ndash103

upon multilocus genotype data We calculated the pro-bability of individual assignments to population clusters(K) without prior information of the origin of individualsA series of tests was conducted using different numbers ofpopulation clusters (maxpops 1ndash30) to guide an empiricalestimate of the number of identifiable populations withburn in and replication values set at 100 000ndash1000 000Each test yielded a log likelihood value of the data (Lnprobability) the highest of which would indicate whichtest was closest to the actual number of genetically distinctpopulations These tests also provided an alpha value themeasure of admixed individuals in the data set Individualswere assigned a probability of assignment to a populationor jointly to two or more populations if their genotypeprofile indicated that they were admixed

fstat 293 was used to calculate Weir amp Cockerhamrsquos(1984) estimator of FST (θ hereafter referred to as FST)and RST values (Slatkin 1995) for each locus and globallyConfidence intervals (99) were calculated for FST valuesto determine if they differed significantly from zero using15 000 bootstrap replicates over loci Pairwise FST valuesfor all pairs of populations were calculated using arle-quin (Schneider et al 2000) Isolation-by-distance wasexamined (with the Mantel test and 1000 permutations)in genepop using the pairwise FST as a measure of geneticdifferentiation Isolation-by-distance was assessed for thecomplete data set and two subsets (1) populations belong-ing to the M caurina clade and (2) mainland populationsbelonging to the M americana clade (excluding introducedisland populations) The Mantel test was used to test forcorrelation between pairwise FST values and Ln geograph-ical distance (in km)

An analysis of molecular variance (amova Excoffieret al 1992) was performed using arlequin (Schneider et al2000) The analysis calculated the partitioning of the mole-cular variance among groups among populations withingroups and among individuals using both allele frequencydata (FST) and allele size and frequency data (RST) Popu-lations were divided into two groups M caurina or Mamericana (see Table 1 for population clade identities)identified by our neighbour-joining dendrogram andpreviously by mitochondrial DNA analysis (Stone amp Cook2002 Stone et al 2002) The two groups were also testedfor differences in allelic richness observed and expectedheterozygosity FST and FIS values using fstat 293

Results

Allele frequencies and heterozygosity

Considerable variability was observed across the 14microsatellite loci (data available upon request) Numbersof alleles per locus ranged from 2 (locus MA15) to 23 (locusMA1) and values of HE ranged from 006 (Admiralty

Island) to 068 (central British Columbia Table 1) Meannumber of alleles was highest in interior Alaska (557) andlowest in Admiralty Island (129) a population that wasfixed for a single allele at most loci Allelic richness washighest in southern Montana (42) and lowest in AdmiraltyIsland (12) Caurina populations had a higher frequencyof unique alleles than americana populations (233 vs 078alleles per population respectively allele frequency dataavailable upon request) and significantly lower HE and HO(Table 1) Allele frequencies differed among populationsand allele size classes and allele frequencies differedbetween the two mtDNA clades (data available upon re-quest) The allele size class divergence was most noticeableat MA1 where allele sizes for the americana clade rangedfrom 207 to 223 bp and allele sizes for the M caurina claderanged from 195 to 205 bp At MVIS020 MER041 MVIS072and MVIS075 caurina populations had a subset of thealleles present in the americana clade Caurina populationswere fixed for allele 195 at MA15 and americana popu-lations had two alleles 195 and 197 with mainly a higherfrequency of allele 197 Kuiu Island and southern Montanapopulations from areas of sympatry between caurinaand americana clades had low frequencies of allele 197 atMA15

HardyndashWeinberg and genotypic equilibrium

Significant FIS values (FIS gt 012 Table 1) in most popula-tions suggest that populations experience some inbreed-ing In locus by population tests excess homozygositycaused departures from HWE at MA5 (Mitkof Island)MA15 (Baranof Kupreanof Mitkof and Prince of WalesIslands) MA14 (Baranof Island) MVIS020 (BaranofChichagof Mitkof Revillagigedo Islands Juneau andsouthern British Columbia) MVIS072 (Kuiu Island) andMVIS075 (Baranof Cleveland Kuiu Kupreanof Prince ofWales Revillagigedo Islands and Thomas Bay) Excessheterozygosity caused departure from HWE at MA11(Cleveland Island) A total of 21 tests of 560 were out ofHWE for homozygote or heterozygote excess with 19departures in americana populations and two in KuiuIsland a mixed caurinaamericana population Tests forgenotypic linkage disequilibrium indicated that MA2 andMA8 were associated in Chichagof and Mitkof Islandpopulations further suggesting that these populationshave experienced some inbreeding

Clustering of individuals

Genetic relationships among individuals determined bythe allele-sharing distance define two groups (Fig 2)Group I contained individuals from mtDNA caurinapopulations and the contact populations with both mtDNAcaurina and mtDNA americana (Kuiu Island and southern

94 M P S M A L L K D S T O N E and J A C O O K


Montana) while Group II contained individuals from allmtDNA americana populations Most americana individualsgrouped in multipopulation clusters In contrast puremtDNA caurina populations (Admiralty Island QueenCharlotte Island Vancouver Island and Oregon) werehighly distinctive with all individuals clustering withintheir respective population groups Kuiu Island formed adistinct cluster but four individuals were within Group IITwo of these were within the Baranof Island cluster andtwo were within a highly mixed cluster (Fig 2 see Table 2for mtDNA haplotypes of Kuiu Island and southernMontana individuals) Most southern Montana individualsclustered with the Oregon samples although one placednear the Admiralty Island population and two were withinGroup II (one within Chichagof Island and one within ahighly mixed cluster)

Phylogenetic analyses and PCA

Some same-sized microsatellite alleles in caurina andamericana may be homoplastic and the result of convergentevolution (Queney et al 2001) because mtDNA analysissuggests a deep separation between these groups (Stone ampCook 2002) Thus we present chord distances in the NJanalysis A NJ assessment of pairwise distances using Neirsquos

infinite allele model revealed the same well-supportednodes as the chord distance tree with minor differencesoccurring at unsupported nodes NJ and ML performedsimilarly and provided a view of the evolutionary historyand geographical structure of these populations Thesedata also provided insight into the history of human-mediated introductions of marten to several islands in thearchipelago The unrooted network of chord distancesamong populations in NJ and ML analyses identified twowell-defined groups supported with bootstrap values of100 (NJ) and 995 (ML) (Fig 3) Group I consisted ofcaurina populations and the two populations with bothamericana and caurina mtDNA haplotypes (Kuiu Islandand southern Montana) identified previously by Stoneet al (2002) Group II contained americana populationsBranch lengths within each group indicated differentevolutionary histories within caurina populations werehighly divergent and separated by large chord distancessuggesting longer periods of isolation for each populationKuiu Island a contact population appeared betweenGroups I and II but was included in Group I with 100support (Fig 3) Human-introduced populations arereflected by the grouping of Baranof Island with its sourcepopulation Thomas Bay and Prince of Wales Island withits source population Cleveland Peninsula (Elkins amp

Fig 2 Unrooted network of genetic relation-ships among 413 American martens (Martesamericana) inferred from allele-sharingdistances constructed using fitch in phylip35c (Felsenstein 1993) All unique genotypesare represented by a line Group I (shaded)includes individuals from caurina-type popu-lations and contact zone populations GroupII (unshaded) includes individuals fromthe americana-type populations and contactzone Population clusters are indicated bysemicircles and labelled with populationnames Individuals from Kuiu Island andsouthern Montana clustering within theamericana clade are labelled (see text)



Nelson 1954 Burris amp McKnight 1973) Several sourceswere used to stock Chichagof Island which has an un-supported central position (Fig 3) Geographic structure ismore evident in Group II where proximal populationsformed well-supported clusters (Mitkof Island and Kup-reanof Island at 84 Revillagigedo Island and Cleveland

Peninsula at 87) Within Group II much shorter branchesdefined most populations (except Baranof and Revilla-gigedo islands and Yakutat)

PCA (figure not shown) was similar to ML and NJanalyses with caurina and americana divided into twodistinct population clusters along the first axis Similarly

Table 2 Bayesian assignments of martens from Kuiu Island and southern Montana using structure (Pritchard et al 2000) IndividualmtDNA identity (a = americana c = caurina) is in mtDNA column The archive number for individualrsquos tissue stored at the Alaska FrozenTissue Collection Fairbanks AK museum is given in the AF column Model was given no prior information of the origin of individualsand asked to assign them to two or 20 clusters based upon their genotype Under two clusters we list the probability of membership inamericana-dominated cluster or caurina-dominated cluster Under 20 clusters we list the probability of each individualrsquos membership in themajority cluster for their population (79 KUIU 57 SMT respectively see Table 4) If membership in the cluster of origin is below 85we list probabilities of membership(s) in other clusters identified by the dominant population(s) in those clusters (abbreviations followTable 1) When individual is assigned membership to undefined clusters (clusters with less than 10 of any single population) we give thenumber of clusters and approximate membership percentage in each

Two clusters 20 clusters Probable membershipin KUIU cluster

20 clusters Probable membershipin other clustermtDNA AF no Individual americana caurina

c 19887 KUIU1 024 076 0888a 19888 KUIU2 0693 0307 0447 KUP MIT 03 CHIC AK 01a 17533 KUIU3 001 099 0947c 17542 KUIU4 0011 0989 0961a 17543 KUIU5 0006 0994 0948c 17544 KUIU6 0004 0996 0967a 17545 KUIU7 0002 0998 0968c 17549 KUIU8 0991 0009 0006 TB BAR 091a 17534 KUIU9 0997 0003 0003 TB BAR 0934c 17552 KUIU10 0308 0692 07 KUP MIT 009 QCI 0045c 17553 KUIU11 0155 0845 0935a 17535 KUIU12 0002 0998 0971a 17536 KUIU13 0101 0899 0945a 17537 KUIU14 003 097 0945a 17538 KUIU15 0226 0774 0893c 17539 KUIU16 0486 0514 0465 KUP MIT 042c 17540 KUIU17 0034 0966 0871c 17541 KUIU18 0355 0645 0818 KUP MIT 0037c 25301 KUIU19 0109 0891 0936a 25310 KUIU20 0025 0975 0931c 25302 KUIU21 0269 0731 091a 25303 KUIU22 0519 0481 0532 KUP MIT 0176 CP POW 0075a 25305 KUIU23 0003 0997 0951c 25306 KUIU24 004 096 0966a 25309 KUIU25 0003 0997 0949

Probable membership in SMT cluster

c 23185 SMT1 0045 0955 0755 7 clusters 0022c 23186 SMT2 0007 0993 0866c 23187 SMT3 099 0009 0005 CHIC AK 034 9 clusters 005c 23192 SMT4 002 0978 0371 ADM 025 KUIU 012c 23180 SMT5 099 0003 0002 CHIC AK 087c 23181 SMT6 002 0977 033 8 clusters 007a 23182 SMT7 0005 0995 0909a 23188 SMT8 0008 0992 0802c 23189 SMT9 0008 0992 0867c 23190 SMT10 0016 0984 0595 ADM 024c 23191 SMT11 0003 0997 0769 8 clusters 002



caurina populations were widely divergent and americanapopulations clustered more tightly Only the first axis wassignificant explaining 324 of the variation (P lt 0001)The first two axes explain cumulatively nearly half of thetotal genetic diversity (436) Populations of caurina wereseparated along the second axis (112 variation not sig-nificant) with Admiralty Island most distinct The americanapopulations separated along the third axis (85 variationnot significant)

Admixture analysis and assignment tests

In the admixture analysis we tested substructuring ofthe data by assuming different numbers of clusters orpopulations ranging from all data belonging to a singlepopulation to the data belonging to 30 populations(Table 3) The probability of the number of populations (K)was estimated in each case (Ln probability of the data)without using any prior population information so thatindividuals were assigned to a cluster based upon theirmultilocus genotype profile The admixture parameter αdetected with each K-value was also estimated Our datawere sampled from 20 locations but the highest pro-bability of the data (Ln = minus13 3978 α = 00291) was foundwith clusters set at 15 However the lowest α value wasfound with populations set at 10 (Ln = minus13 4897 α =0007) This suggests the number of clusters defined byallele frequencies is lower than 20 but greater than 10With more than 10 clusters some individuals were more

difficult to assign rather than more admixed Theseresults support the population and allele-sharing analysesindicating several well-defined populations and somerather indistinct populations

In the admixture analysis individual assignmentsto clusters indicated which population allele frequencyprofile was closest to the individualrsquos genotypic profile

Fig 3 An unrooted network of chorddistances among 20 American marten(Martes americana) populations inferredfrom the neighbour-joining analysisAbbreviations include lsquoClevelandrsquo forlsquoCleveland Peninsula lsquoIsrsquo for island andlsquoInterior AKrsquo for Yukon Flats Alaska andall others follow Table 1 Bootstrap valuesabove 60 are placed at the nodes Boot-strap values derived from a consensus treerepresent the percentage of 1000 trees wherepopulations beyond the node groupedtogether Group I included caurina popu-lations and mixed caurinaamericana popu-lations and Group II included americanapopulations

Table 3 Bayesian clustering analyses for pooled Martes americanadata 413 individuals analysed at 14 microsatellite loci wereassigned to clusters using structure (Pritchard et al 2000) with-out using prior information of population origin The number ofclusters is indicated by K Probability of the data is in the Ln prob-ability of K clusters column and the variance of the probability ispresented in the variance Ln column Alpha values indicate theadmixture value Lowest K and alpha values are in bold type

KLn probability of K clusters

Variance Ln Alpha

1 minus17 6971 657 69332 minus15 4749 2380 00355 minus14 3590 6670 003210 minus13 4883 12570 000712 minus13 4935 13426 003015 minusminusminusminus13 3978 14345 002920 minus13 6655 19820 002822 minus13 7219 21142 002825 minus13 6755 15203 002830 minus13 6414 16670 0028



Admixed individuals were identified when they wereassigned to clusters other than the cluster containing mem-bers of their original population or assigned to multipleclusters If total populations was set at two all americanaindividuals were assigned to one cluster while all caurinaindividuals were assigned to the other cluster with theexception of the contact populations Kuiu Island andsouthern Montana In those populations 23 and 19respectively were assigned to the americana cluster ForKuiu Island and southern Montana individuals we showtheir mitochondrial identification (Stone amp Cook 2002Stone et al 2002) and microsatellite profile and their popu-lation assignment based upon microsatellite genotype(Table 2) Several individuals from Kuiu Island showedevidence of various levels of admixture with combinationsof microsatellite profiles (membership in both caurinaand americana clusters) while all southern Montana indi-viduals had either caurina or americana microsatellite pro-file types (gt 95 membership in either cluster)

The assignment test was conducted without prior infor-mation of the population origin of individuals with clus-ters set at 20 Eleven populations had 80 or more of theirmembership assigned to a single cluster with some of theseclusters containing members from multiple populations(Table 4) The other populations had members distributed

among the clusters In Kuiu Island two members wereassigned at least 30 membership in the Mitkof IslandKupreanof Island cluster and two assigned at least 90membership to the Baranof Island Thomas Bay cluster(Table 2 see 20 clusters column) In southern Montana twoindividuals had highest membership (34 and 87) in theChichagof Island interior Alaska cluster and five indi-viduals had partial membership in several clusters (Table 2see 20 clusters column)

AMOVA analysis FST and RST

When populations were divided into caurina andamericana groups the amova tests indicated that asubstantial portion of the microsatellite genetic variabilityexists between the clades (Table 5 FST = 165 and RST= 33) Microsatellite genetic structure thus parallelsmtDNA clade identification as suggested by other analysesin this study The RST value was more than twice the FSTvalue indicating that caurina and americana clades differin allele size distributions as well as allele frequenciesOverall FST values (Table 1) show high divergence amongcaurina populations (042) and moderate divergence amongamericana populations (014) The two clades also differsignificantly in allelic richness and observed and expected

Table 4 Bayesian assignment test results Proportion of membership (above 001) of each predefined population in each of the 20 clustersgiven no prior information of population origin using structure (Pritchard et al 2000) Program used 100 000 burnins and repetitionsHighest proportion of membership assigned to original population is in bold type Population (Pop) abbreviations follow abbreviationsin Table 1

Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

ADM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 097 0BAR 0 0 0 001 001 0 0 0 0 004 001 0 0 0 0 001 0 090 000 0NBC 001 001 001 019 007 001 001 001 002 022 022 002 001 001 001 012 001 002 006 001CBC 001 001 001 011 013 001 001 001 008 021 015 008 001 001 002 009 001 006 001 001CHIC 0 0 0 001 001 001 0 0 001 085 002 001 0 0 0 003 0 002 001 0CP 001 001 001 002 038 001 001 001 014 011 007 017 001 001 001 003 001 002 0 001QCI 0 097 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0JUN 0 0 0 002 001 0 0 0 002 007 081 001 0 0 001 001 0 001 0 0KUIU 0 001 0 001 001 079 0 0 0 001 0 0 0 0 001 005 0 008 001 0KUP 0 001 0 001 001 001 0 0 001 002 001 0 0 0 0 090 0 001 0 0MIT 0 0 0 003 005 0 0 0 001 002 003 003 0 0 0 073 0 006 0 0SMT 002 002 002 001 002 002 002 002 0 011 001 001 002 002 057 001 002 0 005 002NMT 001 001 001 016 007 001 001 001 003 011 030 004 001 001 001 012 001 002 003 001OR 0 001 0 0 0 0 0 0 0 0 0 0 0 0 093 0 0 0 0 0POW 0 001 0 001 079 0 0 0 001 006 004 001 0 0 0 004 0 001 0 0REV 0 0 0 001 001 0 0 0 094 001 0 001 0 0 0 001 0 001 0 0TB 0 0 0 002 006 001 0 0 004 010 002 005 0 0 0 004 0 062 001 0VAN 0 001 0 0 0 0 0 0 0 0 0 0 0 0 095 0 0 0 0 0YAK 0 0 0 086 005 0 0 0 001 001 002 001 0 0 0 001 0 001 001 0AK 001 0 0 005 004 001 0 0 002 035 037 002 0 0 001 007 0 004 001 0



heterozygosity (Table 1) The caurina clade had signi-ficantly lower allelic richness (P lt 005) and heterozygos-ity (P lt 001) and significantly higher genetic subdivision(P lt 001) than the americana clade in sum suggestinglonger isolation times of caurina populations and ongoinggene flow and more recent expansion of americana intosoutheastern Alaska

Pairwise FST

Pairwise FST values (Table 6) showed varying populationdifferentiation with 175190 comparisons significantlygreater than zero Pairwise FST values in caurina popu-lation comparisons were mainly above 03 indicatinglittle overlap in allele frequencies and great divergence afinding also reflected by the NJ tree Lower pairwise FSTvalues in americana population comparisons indicatemore overlap in allele frequencies also seen in the distancerelationships and lack of bootstrap support for somenodes in the NJ tree Northern British Columbia wasundifferentiated from other americana populations in 11comparisons probably a result of small sample size (n = 5)Northern Montana was undifferentiated from otherpopulations in four comparisons also due possibly tosmall sample size (n = 11) or to gene flow with southernMontana The Mantel test for isolation-by-distanceindicated that genetic and geographical distances wereindependent (P gt 005) even when the two mtDNA cladeswere analysed separately In many cases within the PacificNorthwest oceanic straits appear to be greater barriers togene flow than distance

Discussion

The advance and retreat of glaciers during the Pleistocenecreated a dynamic history of isolation fragmentation andsecondary contact among forest-associated species in the

Pacific Northwest (Pielou 1991 Hewitt 1996) A number ofspecies represented by deeply diverged lineages haveresponded independently to this dynamic geologicalhistory (eg Cook et al 2001) Genetic structure in coastalmarten populations reflects a very distinctive history fromthat found in continental populations Microsatellite datain conjunction with mtDNA and morphological datasuggest that coastal caurina populations have been isolatedfor extended periods M caurina individuals may have colon-ized northward along the coast following the establish-ment of mature forests in this region Coastal islands thenwere formed along the continental shelf when landmasseswere isolated by rising sea level in the early Holoceneand caurina populations may have been fragmented andisolated In contrast americana individuals appear to bestill expanding out of Pleistocene-delimited ranges andnow are contacting and hybridizing with caurina indi-viduals Although the rapid evolution of microsatellitesmay have created homoplasy when comparing allelesacross these distinctive clades (Garza amp Freimer 1996) thisnuclear analysis parallels results from mtDNA and morpho-logical analyses This new perspective on fine-scale gen-etic structure within caurina and americana populationsallowed further assessment of hybridization in addi-tion to providing insight into distinctive colonizationpatterns of caurina and americana populations in the PacificNorthwest

Nuclear vs mitochondrial perspectives

Concordance between nuclear and mtDNA perspectives inmarten contrasts with phylogeographical patterns foundin coastal brown bears another carnivore inhabiting theislands of southeastern Alaska (Paetkau et al 1998)Paetkau et al (1998) demonstrate tight connectivity amongAdmiralty Baranof and Chichagof Island populations andnearby coastal mainland and interior populations of brown

Table 5 amova results for Martes americana data using arlequin (Schneider et al 2000) Populations were assigned into two groupsidentified by the neighbour-joining analysis and clade association (americana and caurina) Fst amova used allele frequency differences asthe distance measure Rst amova used the sum of the squared allele size difference as the distance measure

Source of variation dfSum of squares

Variance components

Percentage of variation

Fixation indices P-value

Fst amova Among groups 1 316139 0840 1649 Fct = 01649 lt 001Among populations within groups 18 726124 0902 177 Fst = 02120 lt 001Within populations 806 2701883 3352 6581 Fsc = 01649 lt 001Total 825 3744145 5094

Rst amova Among groups 1 67441523 190454 3301 Fct = 03301 lt 001Among populations within groups 18 91142744 116879 2026 Fst = 05326 lt 001Within populations 806 217354854 269671 4674 Fsc = 03024 lt 001Total 825 375939121 577004



bears using nuclear microsatellites However that nuclearpattern differs from the highly distinctive insular mito-chondrial lineages identified only in the bears fromAdmiralty Baranof and Chichagof islands Scandinavianbrown bears (Waits et al 2000) show a similar pattern ofstrikingly different nuclear and mitochondrial structureMale-mediated gene flow is suggested as the cause for thediscordant nuclear and mitochondrial patterns in bothstudies Differing life history characteristics between maleand female brown bears such as larger home ranges andgreater dispersal from natal range in males (Canfield ampHarting 1987) and different Ne between microsatellite andmitochondrial loci apparently impact patterns of variationin nuclear vs mitochondrial DNA In contrast nuclearstructure in martens paralleled mitochondrial structureindicating that dispersal of both males and females isinhibited among most islands and between the mainlandand most islands

From a conservation perspective Moritz (1994) suggeststhat evolutionarily significant units should not only bereciprocally monophyletic for mtDNA alleles but alsomaintain significantly diverse allele frequencies at nuclearloci Others (Crandall et al 2000) emphasize the need toconsider adaptive diversity within a species when con-sidering evolutionary processes in conservation biologyWe identified significant partitioning between clades bothin allele frequencies and allele sizes at 14 microsatelliteloci corroborating the patterns in the mitochondrialcytochrome b gene (Carr amp Hicks 1997 Stone et al 2002)the nuclear aldolase C gene (Stone amp Cook 2002) andearlier morphological work (Merriam 1890 Anderson 1970Giannico amp Nagorsen 1989) Because caurina populationshave historically occupied coastal habitats and americanapopulations have occupied inland habitats genetic andmorphological patterns may reflect adaptations associatedwith different habitats We suggest that these forms may

Table 6 Weir and Cockerhamrsquos (1984) Fst for all pairs of marten populations calculated using arlequin (Schneider et al 2000) Populationabbreviations follow Table 1

ADM BAR NBC CBC CHIC CP QCI JUN KUIU KUP

ADM 0000BAR 0632 0000NBC 0653 0214 0000CBC 0531 0158 0001 0000CHIC 0582 0249 0094 0077 0000CP 0528 0206 0041 0045 0091 0000QCI 0756 0499 0438 0358 0470 0380 0000JUN 0554 0232 0054 0056 0130 0079 0413 0000KUIU 0496 0367 0217 0233 0294 0268 0377 0301 0000KUP 0545 0223 0095 0115 0132 0120 0426 0175 0281 0000MIT 0557 0213 0059 0080 0156 0082 0426 0128 0288 0053SMT 0543 0307 0127 0117 0228 0158 0298 0189 0213 0212NMT 0611 0198 0029 0023 0116 0058 0393 0039 0277 0109OR 0610 0446 0318 0280 0371 0323 0440 0340 0332 0360POW 0543 0220 0070 0058 0152 0047 0371 0133 0292 0150REV 0646 0309 0208 0158 0191 0117 0513 0179 0405 0231TB 0521 0148 0076 0091 0123 0087 0428 0142 0252 0140VAN 0620 0480 0383 0332 0400 0353 0484 0375 0339 0404YAK 0598 0272 0096 0131 0237 0185 0460 0143 0354 0235AK 0533 0206 minus0004 0037 0079 0081 0389 0037 0254 0127

MIT SMT NMT OR POW REV TB VAN YAK AKMIT 0000SMT 0187 0000NMT 0110 0155 0000OR 0347 0117 0325 0000POW 0123 0166 0083 0333 0000REV 0245 0335 0162 0449 0214 0000TB 0105 0187 0139 0345 0122 0196 0000VAN 0391 0283 0376 0369 0395 0485 0358 0000YAK 0233 0253 0110 0398 0195 0269 0208 0452 0000AK 0103 0163 0029 0334 0103 0203 0137 0357 0114 0000

Not significantly different from zero (α = 005)



well represent distinctive species and that further invest-igations of ecological behavioural or physiological charac-teristics (eg Aune amp Schladweiler 1997 Ben-David et al1997) should be completed within this emerging frame-work Investigations centred on the two contact zones ofmarten provide opportunities to explore genetic differ-ences involved in speciation and the potential exchange ofgenes between caurina and americana individuals

During the mid-1900s human-mediated introductionsof martens took place on Chichagof Baranof and Prince ofWales Islands where martens were presumed not to exist(Elkins amp Nelson 1954 Burris amp McKnight 1973 MacDon-ald amp Cook 1996) Microsatellite signatures indicated noprior inhabitation of these islands by caurina individuals afinding consistent with mtDNA surveys (Stone et al 2002)High diversity and rare alleles in these introduced popula-tions probably reflect that the founders consisted of severalamericana individuals or that multiple introductions occurredIf caurina individuals had been present on these islandshuman-mediated transplants of americana individuals tocaurina populations may have resulted in genetic swamp-ing of caurina populations such as might be in progresscurrently on Kuiu Island These introductions occurredwithout prior knowledge of the considerable geneticstructure found within species or the potential negativeramifications of hybridization (for examples see Rhymeramp Simberloff 1996) Future management of these furbearerpopulations should reflect this new framework

Hybridization

Hybrid zones provide opportunities to measure geneexchange between diverging taxa (Barton amp Hewitt 1985)On the basis of morphological characteristics Wright(1953) reported hybridization between individuals ofthe American marten clades in western Montana Micro-satellite analysis provided further information on hybrid-ization between caurina and americana individuals in thecontact zones of southern Montana and Kuiu Island thesezones were defined previously by cytochrome b haplotypes(Stone et al 2002) In the Bayesian analysis (Table 2) thetwo individuals in the southern Montana population withamericana-type mtDNA profiles were assigned membershipin the caurina microsatellite cluster while two individualswith caurina-type mtDNA were assigned to the americanamicrosatellite cluster This suggests that hybridizationhas occurred within the southern Montana populationas indicated previously by Wright (1953) who reportedskulls of intermediate size between caurina and americanaindividuals in western Montana

On Kuiu Island both hybridization and introgressionhave occurred between caurina and americana individuals(Table 2) Half of Kuiu Island individuals possessedamericana haplotypes that were common to nearby island

(Mitkof and Kupreanof Islands) and mainland popula-tions The remaining caurina haplotype was unique to KuiuIsland (Stone amp Cook 2002) Thus americana haplotypesappear to be recent arrivals from nearby islands while thedistinctive caurina haplotype is endemic to Kuiu IslandTwelve martens with americana-type mtDNA were assigned30ndash99 membership in the caurina microsatellite clusterand eight martens with caurina-type mtDNA were assigned10ndash99 membership in the americana microsatellite cluster(Table 2) supporting the hypothesis that caurina andamericana individuals are hybridizing on this islandMicrosatellite alleles from mainland americana populationsmay be entering the Kuiu Island nuclear gene pool viaimmigrant movement across Mitkof and Kupreanof IslandsBecause martens on Kuiu Island are hybrids characterizedby either americana or caurina mtDNA haplotypes witha mixture of nuclear genotypes both male and femaleamericana individuals appear to be expanding into thisarea We suspect that Kuiu Island was inhabited previ-ously solely by caurina individuals However americanaindividuals may be competitively superior such that popu-lations of caurina in coastal regions may be persisting onlyon islands isolated by substantial oceanic straits If caurinapopulations existed on the mainland or near-shore islandsthey may have been vulnerable to competition by americanaindividuals when this group colonized the region

The zone of contact in southeastern Alaska Kuiu Islandis the furthest of a string of three islands that extend fromthe mainland (Fig 1) Very narrow stretches of waterseparate Mitkof Kupreanof and Kuiu Islands effectivelycreating a peninsular effect (MacDonald amp Cook 1996)Individuals from mainland americana populations pro-bably colonized Mitkof Kupreanof and Kuiu Islandsacross these narrow oceanic straits The original specimentaken from southeastern Alaska in 1909 was collected fromKuiu Island (Swarth 1911) and identified as the caurina morphBoth morphs are present in southeastern Alaska with theamericana clade distributed across a larger range (due par-tially to human introductions to several large islands) whilethe caurina clade is restricted to Admiralty and Kuiu Islands

Colonization and evolutionary history

Microsatellite data illuminated fine-scale relationshipsamong populations including genetic signatures related toextended periods of isolation and recent expansions Allanalyses are consistent with a longer period of isolationfor caurina populations when compared to the natural (ienonintroduced) americana populations Networks based onallele-sharing distances (pairwise individual comparisons)and assignment tests show similar patterns suggesting pastisolation of caurina populations and possible expansionandor extensive gene flow among americana popula-tions Further significantly lower allelic richness and



heterozygosity and significantly higher division amongpopulations characterized caurina in contrast to americanaHigher frequency of private alleles and HardyndashWeinbergequilibrium in caurina populations indicates that althoughthese populations have been isolated for an extended timethey are large and stable enough to maintain low-frequencyalleles (Queney et al 2001) and avoid inbreeding In contrastamericana populations are modestly divergent have higherallelic richness and tend to depart from HWE suggestingmore recent expansion into the Pacific Northwest largereffective population sizes and possibly continuing geneflow In sum our results illustrate distinctive evolutionaryhistories for the coastal caurina and continental americanaforms of marten

The monomorphic genetic pattern in the AdmiraltyIsland caurina population may have arisen due to repeatedbottlenecks (Eldridge et al 1999) or from a small effectivepopulation size over an extended time (Avise et al 1984)The Admiralty Island population is unlikely to have beenfounded recently by human introduction due to its uniquecytochrome b haplotype (Stone amp Cook 2002) and micro-satellite profiles Other documented introductions ofamericana individuals resulted in genetically diverse popu-lations In contrast to the brown bear studies (eg Paetkauet al 1998) nuclear and mitochondrial perspectives onAdmiralty Island marten are more similar to geneticpatterns found in Prince of Wales Island flying squirrels(Bidlack amp Cook 2002) Oceanic straits beyond a certainwidth may thus serve as barriers to gene flow for mid-sizedand smaller mammals such as marten and flying squirrelsbut not for large mammals such as bears

Genetic distance among marten populations was inde-pendent of geographical distance even when analysedwithin clades Structural patterns are likely confounded bythe landscape highly fragmented both in time and spacein which populations are located Island populations ofcaurina may have colonized by expanding northwardalong the coast from a southern refuge roughly 10 000ndash12 000 years ago after glaciers receded from the coast andas forest habitat was becoming established Rising sea levelmay have isolated founders on Admiralty Kuiu QueenCharlotte and Vancouver Islands around 10 000 years ago(Warner et al 1982 Pielou 1991 Fedje amp Josenhans 2000)In contrast americana populations may be still in theprocess of expanding from continental refugia using rivercorridors through the coastal mountain ranges to accessthe Pacific coast as these corridors became accessible withthe recession of glaciers Hybridization and introgressionon Kuiu Island may represent the furthest front of naturalamericana population expansion into the islands supportinga hypothesis of recent expansion into southeastern AlaskaContinued gene flow among the expanding americanapopulations and a shorter time since colonization maypreclude divergence into distinct populations

Kyle et al (2000) studied northern continental popula-tions of americana and concluded that few barriers togene flow exist among marten populations from theYukon through central Northwest Territories Canada Theyattributed the lack of genetic structure to isolation-by-distance and predicted higher population fragmentationshould be found in southerly regions where habitat wasmore disjunct Indeed we found that population geneticstructure in caurina and to a lesser extent in americanacould be attributed to fragmentation However in contrastamericana population structure in this region was not dueto isolation-by-distance and was less fragmented by barrierssuch as the Coast Mountain Range that parallels coastalsoutheastern Alaska Oceanic straits may prevent geneflow among caurina populations and may also disrupt geneflow among americana populations Revillagigedo Islanda near-shore island was the most distinctive americanapopulation Extensive timber harvesting may also befragmenting populations of this forest-associated speciesThe persistence of caurina populations on larger islandsin southeastern Alaska suggests that if habitat islands ofsimilar size are preserved these insular marten popula-tions can persist However we do not know whethercaurina populations inhabited Alaska islands other thanAdmiralty and Kuiu Islands in the past and have subse-quently gone extinct New palaeontologic work across thearchipelago may shed light upon this (eg Heaton amp Grady2003) The americana individuals appear to disperse betterand outcompete caurina individuals when they contactThus to maintain caurina populations in southeasternAlaska caurina populations may need habitat inaccessibleto americana individuals

Conclusions

Microsatellite genetic structure corroborated the divisionbetween mitochondrial clades found in American martenwith a pronounced break between Martes caurina andM americana populations Population structure withineach clade indicated different evolutionary historiesfor the coastal M caurina and continental M americanapopulations Hybridization between individuals fromthese distinctive morphs was identified in Kuiu Islandand in southern Montana In light of mtDNA evidencemicrosatellite analysis suggests that M caurina populationsin southeastern Alaska were founded soon after glacialretreat and then isolated by rising sea levels In contrastindividuals of the M americana clade appear to beexpanding into southeastern Alaska and hybridizing withM caurina individuals where they contact The dynamicsand viability of the insular M caurina populations inparticular should be carefully monitored Given thetremendous habitat modification that has occurred inPacific Northwest forests over the past century (eg Durbin



1999) thoughtful management of these ecosystems will berequired Decisions should be based on the realization thatwe still have an incomplete taxonomic and evolutionaryframework for a significant portion of the biotic diversityof this complex region including much of the purportedlywell-studied mammalian fauna

Acknowledgements

We extend our appreciation to John Demboski Merav Ben-DavidSteve MacDonald Enrique Lessa and Rod Flynn for ideas andcomments to Kyndall Hildebrandt for assistance in data collec-tion to Corey Davis and Curtis Strobeck for the early release ofmicrosatellite primers to Tommy LeCroy and Erin OrsquoLeary-Jepsen for assistance in the DNA Core Laboratory and to the manyprogrammers for shareware available via the internet We alsothank the following individuals for their contributions to theAFTC N Anderson (Montana Fish Wildlife and Parks) R Flynn(Alaska Department of Fish and Game) R Green (Oregon Depart-ment of Fish and Wildlife) R Marshall M McAdie and T Smith(British Columbia Environment) and the many anonymoustrappers This research was funded by the United States Fish andWildlife Service United States Department of Agriculture ForestService Alaska Cooperative Fish and Wildlife Research UnitNational Science Foundation (DEB0196095 and DEB9981915)National Institutes of Health NCRR 1P20RR016448-01 and theUniversity of Alaska Graduate School Natural Resource GraduateFellowship and Thesis Completion Fellowship and was aidedby a Grant-in-Aid of Research from the National Academy ofSciences through Sigma Xi The Scientific Research Society

References

Anderson E (1970) Quaternary evolution of the genus Martes(Carnivora Mustelidae) Acta Zoologica Fennica 130 1ndash132

Aune KE Schladweiler P (1997) Age sex structure and fecundity ofthe American marten in Montana In Martes Taxonomy EcologyTechniques and Management (eds Proulx G Bryant HN WoodardPM) pp 61ndash77 The Provincial Museum of Alberta Edmonton

Avise JC Neigel JE Arnold J (1984) Demographic influences onmitochondrial DNA lineage survivorship in animal popula-tions Journal of Molecular Evolution 20 99ndash105

Barton NH Hewitt GM (1985) Analysis of hybrid zones AnnualReview of Ecology and Systematics 16 113ndash148

Ben-David M Flynn RW Schell DM (1997) Annual and seasonalchanges in diets of martens evidence from stable isotopeanalysis Oecologia 111 280ndash291

Bidlack AL Cook JA (2002) A nuclear perspective on endemism innorthern flying squirrels (Glaucomys sabrinus) of the AlexanderArchipelago Alaska Conservation Genetics 3 247ndash259

Bowcock AM Ruiz-Linares A Tomfohrde J Minch E Kidd JRCavalli-Sforza LL (1994) High resolution of human evolutionarytrees with polymorphic microsatellites Nature 368 455ndash457

Burris OE McKnight DE (1973) Game transplants in AlaskaAlaska Department of Fish and Game Wildlife Technical Bulletin 41ndash57

Canfield J Harting AL (1987) Home range and movements InGrizzly Bear Compendium (eds LeFranc MN Moss MB Patnode KASugg WC) pp 27ndash35 National Wildlife Federation WashingtonDistrict of Columbia

Carr SM Hicks SA (1997) Are there two species of marten in NorthAmerica Genetic and evolutionary relationships withinMartes In Martes Taxonomy Ecology Techniques and Management(eds Proulx G Bryant HN Woodard PM) pp 15ndash28 TheProvincial Museum of Alberta Edmonton

Cavalli-Sforza LL Edwards AWF (1967) Phylogenetic analysismodels and estimation procedures American Journal of HumanGenetics 19 233ndash257

Clark TW Anderson E Douglas C Strickland M (1987) Martesamericana Mammalian Species 289 1ndash8

Conroy CJ Cook JA (2000) Phylogeography of a post-glacialcolonizer Microtus longicaudus (Muridae Rodentia) MolecularEcology 9 165ndash175

Cook JA Bidlack AL Conroy CJ et al (2001) A phylogeographicperspective on endemism in the Alexander ArchipelagoBiological Conservation 97 215ndash227

Cook JA MacDonald SO (2001) Should endemism be a focus ofconservation efforts along the North Pacific Coast of NorthAmerica Biological Conservation 97 207ndash213

Crandall KA Bininda-Emonds ORP Mace GM Wayne RK (2000)Considering evolutionary process in conservation biologyTrends in Ecology and Evolution 15 290ndash295

Crozier RH (1992) Genetic diversity and the agony of choiceBiological Conservation 61 11ndash15

Davis CS Strobeck C (1998) Isolation variability and cross-speciesamplification of polymorphic microsatellite loci in the familyMustelidae Molecular Ecology 7 1771ndash1788

Demboski JR Stone KD Cook JA (1999) Further perspectives onthe Haida Gwaii glacial refugium Evolution 53 2008ndash2012

Durbin K (1999) Tongass Pulp Politics and the Fight for the AlaskaRain Forest Oregon State University Press Corvallis Oregon

Eldridge MD King BJM Loupis AK et al (1999) Unprecedentedlow levels of genetic variation and inbreeding depression in anisland population of the black-footed rock-wallaby ConservationBiology 13 531ndash541

Elkins WA Nelson UC (1954) Wildlife introductions and trans-plants in Alaska Proceedings of the Fifth Alaska Science ConferenceAnchorage Alaska USA

Excoffier L Smouse PE Quattro JM (1992) Analysis of molecularvariance inferred from metric distances among DNA haplo-types application to human mitochondrial DNA restrictiondata Genetics 131 479ndash491

Faith DP (1992) Conservation evaluation and phylogenetic diver-sity Biological Conservation 61 1ndash10

Fedje DW Josenhans H (2000) Drowned forests and archaeologyon the continental shelf of British Columbia Canada Geology28 99ndash102

Felsenstein J (1993) Phylogeny Inference Package (PHYLIP) Version35c University of Washington Seattle

Fitch WM Margoliash E (1967) Construction of phylogenetictrees Science 155 279ndash284

Fleming MA Cook JA (2002) Phylogeography of endemic ermine(Mustela erminea) in Southeast Alaska Molecular Ecology 11 795ndash808

Fleming M Ostrander EA Cook JA (1999) Microsatellite markersfor American mink (Mustela vison) and ermine (Mustela erminea)Molecular Ecology 8 1352ndash1354

Garza JC Freimer NB (1996) Homoplasy for size at microsatelliteloci in humans and chimpanzees Genome Research 6 211ndash217

Giannico GR Nagorsen DW (1989) Geographic and sexual variationin the skull of Pacific coast marten (Martes americana) CanadianJournal of Zoology 67 1386ndash1393



Goudet J (2001) FSTAT a Program to Estimate and Test Gene Diversitiesand Fixation Indices Version 293 updated from Goudet 1995Available from httpwwwunilchizeasoftwaresfstathtml

Guo SW Thompson EA (1992) Performing the exact test of HardyndashWeinberg proportion for multiple alleles Biometrics 48 361ndash372

Hall ER (1981) The Mammals of North America 2nd edn John Wileyand Sons New York

Heaton TH Grady F (2003) The Late Wisconsin vertebrate historyof Prince of Wales Island southeastern Alaska In VertebratePaleontology of Late Cenozoic Cave Deposits in North America (edsSchubert BW Mead JI Graham R) (in press) Indiana UniversityPress Indiana

Hewitt GM (1996) Some genetic consequences of ice ages and theirrole in divergence and speciation Biological Journal of the LinneanSociety 58 247ndash276

Klein DR (1965) Postglacial distribution patterns of mammals inthe southern coastal regions of Alaska Journal of the Arctic Insti-tute of North America 18 7ndash20

Kyle CJ Davis CS Strobeck C (2000) Microsatellite analysis ofNorth American pine marten (Martes americana) populationsfrom the Yukon and Northwest Territories Canadian Journal ofZoology 78 1150ndash1157

Louis EJ Dempster ER (1987) An exact test for HardyndashWeinbergand multiple alleles Biometrics 43 805ndash811

MacDonald SO Cook JA (1996) The land mammal fauna of South-eastern Alaska Canadian Field-Naturalist 110 571ndash598

McCauley DE (1993) Genetic consequences of extinction andrecolonization in fragmented habitats In Biotic Interactionsand Global Change (eds Kareiva PM Kingsolver JB Huey RB)pp 217ndash233 Sinauer Associates Sunderland MA

McGowan C Howes LA Davidson WS (1999) Genetic analysis ofan endangered pine marten (Martes americana) population fromNewfoundland using randomly amplified polymorphic DNAmarkers Canadian Journal of Zoology 77 661ndash666

Merriam CH (1890) Description of a new marten (Mustela caurina)from the north-west coast region of the United States NorthAmerican Fauna 4 27ndash29

Mitton JB Raphael MG (1990) Genetic variation in the martenMartes americana Journal of Mammalogy 71 195ndash197

Moritz C (1994) Defining lsquoevolutionarily significant unitsrsquo forconservation Trends in Ecology and Evolution 9 373ndash375

Nei M (1972) Genetic distance between populations AmericanNaturalist 106 283ndash292

Paetkau D Shields GF Strobeck C (1998) Gene flow betweeninsular coastal and interior populations of brown bears inAlaska Molecular Ecology 7 1283ndash1292

Pielou EC (1991) After the Ice Age the Return of Life to Glaciated NorthAmerica University of Chicago Press Chicago IL

Pritchard JK Stephens M Donnelly P (2000) Inference of populationstructure using multilocus genotype data Genetics 155 945ndash959

Queney G Ferrand N Weiss S Mougel F Monnerot M (2001)Stationary distributions of microsatellite loci between divergentpopulation groups of the European rabbit (Oryctolagus cuniculus)Molecular Biology and Evolution 18 2169ndash2178

Raymond M Rousset F (1995) GENEPOP Version 12 populationgenetics software for exact tests and ecumenicism Journal ofHeredity 86 248ndash249

Rhymer JM Simberloff D (1996) Extinction by hybridization andintrogression Annual Review of Ecology and Systematics 27 83ndash109

Ryder OA (1986) Species conservation and systematics thedilemma of subspecies Trends in Ecology and Evolution 1 9ndash10

Schneider S Roessli D Excoffier L (2000) ARLEQUIN Version 2001a Software for Population Genetics Data Analysis Genetics andBiometry Laboratory University of Geneva Switzerland

Slatkin M (1995) A measure of population subdivision based onmicrosatellite allele frequency Genetics 139 457ndash462

Stone KD Cook JA (2000) Phylogeography of black bears (Ursusamericanus) of the Pacific Northwest Canadian Journal of Zoology78 1ndash6

Stone KD Cook JA (2002) Molecular evolution of the holarcticgenus Martes Molecular Phylogenetics and Evolution 24 169ndash179

Stone KD Flynn RW Cook JA (2002) Post-glacial colonization ofnorthwestern North America by the forest associated Americanmarten (Martes americana) Molecular Ecology 11 2049ndash2063

Strickland MA Douglas CW (1987) Marten In Wild FurbearerManagement and Conservation in North America (eds Novak MBaker JA Obbard ME Malloch B) pp 531ndash546 Ontario Ministryof Natural Resources Toronto

Swarth HS (1911) Birds and mammals of the 1909 AlexanderAlaska expedition University of California Publications in Zoology7 9ndash172

Swarth HS (1936) Origins of the fauna of the Sitkan DistrictAlaska Proceedings of the California Academy of Science 22359ndash78

Talbot SL Shields GF (1996) Phylogeography of brown bears(Ursus arctos) of Alaska and paraphyly within the UrsidaeMolecular Phylogenetics and Evolution 5 477ndash494

Thompson ID (1994) Marten populations in uncut and logged borealforests in Ontario Journal of Wildlife Management 58 272ndash280

Vane-Wright RI Humphries CJ Williams PH (1991) What toprotect mdash systematics and the agony of choice Biological Con-servation 55 235ndash254

Waits L Taberlet P Swenson JE Sandegren F Franzen R (2000)Nuclear DNA microsatellite analysis of genetic diversityand gene flow in the Scandinavian brown bear (Ursus arctos)Molecular Ecology 9 421ndash431

Warner BG Mathewes RW Clague JJ (1982) Ice-free conditions onthe Queen Charlotte Islands British Columbia at the height ofLate Wisconsin glaciation Science 218 675ndash677

Weir BS Cockerham CC (1984) Estimating F-statistics for theanalysis of population structure Evolution 38 1358ndash1370

Wilson DE Reeder DM eds (1993) Mammal Species of the WorldA Taxonomic and Geographic Reference 2nd edn SmithsonianInstitution Press Washington DC

Wright PL (1953) Intergradation between Martes americana andMartes caurina in western Montana Journal of Mammalogy 3470ndash87

Youngman PM (1993) The Pleistocene small carnivores of easternBeringia Canadian Field-Naturalist 107 139ndash163

Maureen Small is a biologist in the Genetics Division at theWashington State Department of Fish and Wildlife This researchwas a collaborative effort involving Smallrsquos postdoctoral project atIdaho State University and Karen Stonersquos doctoral dissertationwork at University of Alaska Fairbanks both conducted underthe direction of Joseph Cook Karen Stone is now an AssistantProfessor of Biology at Southern Oregon University and Joseph Cookis a Professor of Biology at Idaho State University Cookrsquos work focuseson how evolutionary and biogeographical histories of organismshave shaped patterns of genetic variation in extant populations

90



Molecular Ecology

12 89ndash103

widespread anthropogenic disturbance (Crozier 1992Ryder 1986 Vane-Wright

et al

1991 Faith 1992 Moritz 1994)Although eight subspecies of

M americana

have beendescribed (Clark

et al

1987) these are traditionally placedinto two morphological groups

caurina

and

americana

(Merriam 1890 Clark

et al

1987) Molecular investigationscorroborate the distinction of

caurina

and

americana

as tworeciprocally monophyletic mitochondrial clades (Carr ampHicks 1997 Stone

et al

2002) These forms appear to havediverged due to isolation in distinct southern glacial refugia(

caurina

populations isolated in western and

americana

populations isolated in eastern United States respectivelyCarr amp Hicks 1997 Stone amp Cook 2002 Stone

et al

2002)because there is little evidence of a high latitude refugiumduring the Wisconsin that was forested (Pielou 1991)Youngman (1993) suggests the persistence of

americana

in Beringia during the Pleistocene but material that heexamined was not dated either stratigraphically or through

isotope analysis and may have been deposited during theHolocene

The two forms are largely allopatric with

caurina

popu-lations documented along the West Coast (California tosoutheastern Alaska) and then eastward to WyomingMontana and Idaho (Fig 1 Wright 1953 Hall 1981 Carr ampHicks 1997 Stone

et al

2002) Currently

americana

popu-lations are more widespread distributed from Montanaand Idaho northward to Alaska and eastward to the Atlan-tic Coast Two zones of contact have been identified oneextending through southern and western Montana and theother on Kuiu Island in southeastern Alaska (Stone

et al

2002 Wright 1953 Stone amp Cook 2002) Although Merriam(1890) describes these groups as distinctive species Wright(1953) reports morphological intergradation betweenthe groups in Montana and suggested that they belongto a single species

M americana

a conclusion reflectedby the current taxonomy (Wilson amp Reeder 1993) Carr amp

Fig 1 Map of sampling locations alsoshowing distribution of mitochondrialclades in American martens (Martesamericana) Light and dark grey shadingrepresents regions inhabited by membersof the americana and caurina clades re-spectively with striped regions indicatingcontact zones Inset map shows the NorthAmerican distribution of martens modifiedfrom Hall (1981)


91


Molecular Ecology

12 89ndash103


caurina

and

americana



caurina

and

americana


et al


et al


et al


et al


M caurina


americana


et al


caurina

and

americana



Sampling


deg

N 113ndash126

deg



n


caurina

and

americana

populations(Stone

et al


n


n




americana

or

caurina


et al


m


m


m

MgCl

2


times


deg


deg


deg

C 20 s at 58

deg

C 5 s at 72

deg


deg

C20 s at 54

deg

C 5 s at 72

deg


deg



et al


2


m


m

MgCl

2


m

MgCl

2


Data analysis

fstat


92



Molecular Ecology

12 89ndash103


F

IS


F

IS


genepop


P


P


α



et al


sharedst


per locus The

fitch

program in

phylip



phylip


gendist


neighbour


seqboot

and analysed in

gendist


neighbour


consense


seqboot contml

and

consense


pcagen












Results




































Variance Ln Alpha










Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References











































































91


Molecular Ecology

12 89ndash103


caurina

and

americana



caurina

and

americana


et al


et al


et al


et al


M caurina


americana


et al


caurina

and

americana



Sampling


deg

N 113ndash126

deg



n


caurina

and

americana

populations(Stone

et al


n


n




americana

or

caurina


et al


m


m


m

MgCl

2


times


deg


deg


deg

C 20 s at 58

deg

C 5 s at 72

deg


deg

C20 s at 54

deg

C 5 s at 72

deg


deg



et al


2


m


m

MgCl

2


m

MgCl

2


Data analysis

fstat


92



Molecular Ecology

12 89ndash103


F

IS


F

IS


genepop


P


P


α



et al


sharedst


per locus The

fitch

program in

phylip



phylip


gendist


neighbour


seqboot

and analysed in

gendist


neighbour


consense


seqboot contml

and

consense


pcagen












Results




































Variance Ln Alpha










Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References










































































92



Molecular Ecology

12 89ndash103


F

IS


F

IS


genepop


P


P


α



et al


sharedst


per locus The

fitch

program in

phylip



phylip


gendist


neighbour


seqboot

and analysed in

gendist


neighbour


consense


seqboot contml

and

consense


pcagen












Results




































Variance Ln Alpha










Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References















































































Results




































Variance Ln Alpha










Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References






































































































Variance Ln Alpha










Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References


















































































Clusters

Pop 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20





Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References













































































Pairwise FST


Discussion







Variance components


















Hybridization













Conclusions





Acknowledgements


References























































































Hybridization













Conclusions





Acknowledgements


References
















































































Conclusions





Acknowledgements


References













































































Acknowledgements


References










































































american marten ( martes americana ) in the pacific ...msb.unm.edu/isles/small_et_al 2003 pacific nw...

Documents