Molecular Phylogenetics and Evolution 100 (2016) 170–182

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/locate /ympev

Revision of widespread red squirrels (genus: Tamiasciurus) highlightsthe complexity of speciation within North American forests

http://dx.doi.org/10.1016/j.ympev.2016.04.0141055-7903/� 2016 Published by Elsevier Inc.

⇑ Corresponding author at: Kansas State University, Division of Biology, 116Ackert Hall, Manhattan, KS 66547, USA.

E-mail addresses: ahope@ksu.edu (A.G. Hope), jmalaney@gmail.com(J.L. Malaney), kayce.bell@gmail.com (K.C. Bell), f.salazarmiralles@gmail.com(F. Salazar-Miralles), aschavez@berkeley.edu (A.S. Chavez), bbarber4@uwyo.edu(B.R. Barber), cookjose@unm.edu (J.A. Cook).

Andrew G. Hope a,⇑, Jason L. Malaney b, Kayce C. Bell c, Fernando Salazar-Miralles c, Andreas S. Chavez d,Brian R. Barber e, Joseph A. Cook c

aDivision of Biology, Kansas State University, Manhattan, KS 66506, USAbDepartment of Biology, Austin Peay State University, Clarksville, TN 37044, USAcDepartment of Biology and Museum of Southwestern Biology, University of New Mexico, Albuquerque, NM 87131, USAdMuseum of Vertebrate Zoology and Department of Integrative Biology, University of California-Berkeley, Berkeley, CA 94720, USAeBiodiversity Institute, University of Wyoming, Laramie, WY 82071, USA

a r t i c l e i n f o

Article history:Received 20 November 2015Revised 7 April 2016Accepted 11 April 2016Available online 12 April 2016

Keywords:Boreal biomeQuaternarySky islandSoutheast AlaskaSpeciation with gene flowSpecies tree estimation

a b s t r a c t

Integration of molecular methods, ecological modeling, and statistical hypothesis testing are increasingour understanding of differentiation within species and phylogenetic relationships among species byrevealing environmental connections to evolutionary processes. Within mammals, novel diversity isbeing discovered and characterized as more complete geographic sampling is coupled with newermulti-disciplinary approaches. North American red squirrels exemplify a forest obligate genus whosespecies are monitored as indicators of forest ecosystem condition, yet phylogenetic relationships reflect-ing evolutionary history within this genus remain tentative. Through testing of competing systematic andniche-based divergence hypotheses, we recognize three species, Tamiasciurus douglasii, T. hudsonicus, andT. fremonti. Our data provide evidence of regional differences in evolutionary dynamics and continentalgradients of complexity that are important both for future management and for investigating multiplepathways that can lead to the formation of new species.

� 2016 Published by Elsevier Inc.

1. Introduction

Understanding how the formation of species is shaped by com-mon biogeographic and evolutionary processes (Hewitt, 2000;Taberlet et al., 1998) is a critical step for recognizing and conserv-ing biodiversity (Riddle et al., 2008; Whittaker et al., 2005). Range-wide evolutionary assessments of species that incorporate bothgenetic and niche-based approaches provide critical perspectiveson regional evolutionary and ecological differences in biodiversity.For instance, many vertebrate species associated with the borealbiome in North America have broad continent-wide distributionscoincident with temperate and northern conifer forests (Arbogastand Kenagy, 2001), but these forests are now experiencingincreased mortality (Allen et al., 2015), and regional declines(Coops andWaring, 2011; Gauthier et al., 2015) will likely continue

to impact vertebrate communities. Geography and past climatehave shaped the assembly of communities according to predictablebiogeographic patterns that often reflect common evolutionaryprocesses (Hampe and Petit, 2005). Northernmost contemporarypopulations frequently constitute a leading edge of expansiondue to climate warming following the Last Glacial Maximum(LGM; �the last 12 kyrs). However, high-latitude populationsmay have also experienced spatial and temporal stability if theypersisted through the Last Glacial period (�130–12 ka) withinone or more northern refugia (Brubaker et al., 2005; Hewitt,1999; Lessa et al., 2003). Mid-latitude forest-associated popula-tions often have multiple distinct lineages distributed longitudi-nally across the continent reflecting repeated isolation in discreteareas south of glacial-phase continental ice-sheets. However, thegenetic signatures of independent histories among these lineagesare often obscured at their distributional edges where they nowoverlap geographically (Atkinson et al., 2007; Hope et al., 2014a).Signatures of these complex histories of isolation and reconnectionare often apparent along contact zones (Swenson and Howard,2005) where dynamics of gene flow and interspecific interactionsresult in increased community complexity (Hewitt, 2000). Finally,peripherally isolated populations in southern and coastal regions

http://crossmark.crossref.org/dialog/?doi=10.1016/j.ympev.2016.04.014&domain=pdfhttp://dx.doi.org/10.1016/j.ympev.2016.04.014mailto:ahope@ksu.edumailto:jmalaney@gmail.commailto:kayce.bell@gmail.commailto:f.salazarmiralles@gmail.commailto:aschavez@berkeley.edumailto:bbarber4@uwyo.edumailto:cookjose@unm.eduhttp://dx.doi.org/10.1016/j.ympev.2016.04.014http://www.sciencedirect.com/science/journal/10557903http://www.elsevier.com/locate/ympevhttps://www.researchgate.net/publication/280742445_On_underestimation_of_global_vulnerability_to_tree_mortality_and_forest_die-off_from_hotter_drought_in_the_Anthropocene?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230218003_Comparative_phylogeography_as_an_integrative_approach_to_historical_biogeography?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230218003_Comparative_phylogeography_as_an_integrative_approach_to_historical_biogeography?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/251252562_Longitudinal_patterns_in_species_richness_and_genetic_diversity_in_European_oaks_and_oak_gallwasps?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/229646915_Beringia_as_a_glacial_refugium_for_boreal_trees_and_shrubs_New_perspectives_from_mapped_pollen_data?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/228418666_Estimating_the_vulnerability_of_fifteen_tree_species_under_changing_climate_in_Northwest_North_America?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/281168755_Boreal_forest_health_and_global_change?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/50212247_Conserving_biodiversity_under_climate_change_The_rear_edge_matters?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227727399_Postglacial_recolonization_of_European_Biota?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227727399_Postglacial_recolonization_of_European_Biota?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/265462716_Multilocus_phylogeography_and_systematic_revision_of_North_American_water_shrews_genus_Sorex?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/10618397_From_The_Cover_Genetic_footprints_of_demographic_expansion_in_North_America_but_not_Amazonia_during_the_Late_Quaternary?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/7541093_Clustering_of_Contact_Zones_Hybrid_Zones_and_Phylogeographic_Breaks_in_North_America?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/7541093_Clustering_of_Contact_Zones_Hybrid_Zones_and_Phylogeographic_Breaks_in_North_America?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230205612_Comparative_Phylogeography_and_Postglacial_Colonization_Routes_in_Europe?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227602248_Conservation_Biogeography_assessment_and_prospect?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/237524480_The_role_of_molecular_genetics_in_sculpting_the_future_of_integrative_biogeography?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/profile/Jason_Malaney?el=1_x_100&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/profile/Joseph_Cook?el=1_x_100&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/profile/Andrew_Hope5?el=1_x_100&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/profile/Brian_Barber2?el=1_x_100&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/profile/Kayce_Bell?el=1_x_100&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182 171

generally occur in patches of discontinuous forest. Peripheral pop-ulations can reflect either long-term isolation (P1 glacial phase;e.g., Cook et al., 2001, 2006) or more recent fragmentation of pre-viously continuous habitat since the LGM (Galbreath et al., 2009).Such isolates often constitute important sources of genetic diver-sity for species (Channell and Lomolino, 2000) with high conserva-tion value (Malaney and Cook, 2013).

A growing body of comparative genetic evidence indicates thatevolutionary processes of diversification on a continental scale arerepeatable both across taxonomic groups and through time, wherecryptic diversity is discovered as additional groups are assessed,resulting in recognition of new species, or minimally, distinct evo-lutionary lineages (Hope et al., 2014a). Given that species descrip-tions are ongoing, even among mammals, current taxonomyfrequently fails to adequately reflect the extent of evolutionarydiversity or relationships among extant lineages (Patterson,2000; Bickford et al., 2007), thus impeding effective managementand conservation of fundamental components of biodiversity(Soltis and Gitzendanner, 1999; Hoberg et al., 2012). New toolsetsare facilitating species discovery based on multi-locus sequencedata (Knowles and Carstens, 2007). Concurrently, new statisticalmethods assess ‘‘species” limits, recognizing taxonomic or system-atic arrangements that hold strongest support for actual relation-ships among distinct biological units (reviewed in Carstens et al.,2013). Both historical and ongoing gene flow between related spe-cies (e.g., Sullivan et al., 2014), however, present additional chal-lenges to molecular systematic assessments.

The evolutionary history of North American red squirrels (genusTamiasciurus Trouessart 1880) reflects the influence of long-termQuaternary climate cycles on continental patterns of diversifica-tion and community change (Arbogast and Kenagy, 2001). Periodicallopatry and divergence has been followed by episodes of recon-nection and gene flow as climate cycled through glacial phases(Chavez et al., 2014). Simultaneously, some red squirrel popula-tions, particularly peripheral ones, likely remained stable and iso-lated through extended glacial cycles. As such, systematicrelationships, species limits, and taxonomy within this groupremain tentative. Thorington and Hoffmann (2005) recognizedthree species: Tamiasciurus douglasii Bachman, 1839, distributedin forests from southern California and the Sierra Nevada north-ward through the Cascade and Coastal ranges; Tamiasciurus hud-sonicus Erxleben, 1777, distributed across most of North Americanorth of Mexico including multiple peripheral populations; andTamiasciurus mearnsi Townsend, 1897, restricted within Baja Cali-fornia Norte, Mexico. Previous mitochondrial DNA (mtDNA)assessments (Arbogast et al., 2001; Barber, 2007) documented sev-eral divergent lineages, recounting a complex biogeographic his-tory. Across these species, T. hudsonicus is paraphyletic withrespect to T. douglasii, while T. mearnsi is minimally divergent fromT. douglasii (Arbogast et al., 2001; Barber, 2007). In the AmericanSouthwest, populations of T. hudsonicus, including those on skyislands, potentially represent a third species (Tamiasciurus fre-monti), based on genetic (Arbogast et al., 2001), behavioral, andmorphological differences (Findley, 1961; Hardy, 1950). Morerecently, Chavez et al. (2014) largely corroborated these relation-ships with nuclear data, although mito-nuclear discord indicatedhistoric introgression between populations of the Pacific North-west. However, to date, no study has explicitly tested the validityof systematic relationships using a phylogenetic approach.

The North American red squirrel complex constitutes an excel-lent system for testing diversification hypotheses because theircomplex biogeographic history includes both physical (insular)and ecogeographic isolation of lineages, paraphyly, and gene flow.We present a multi-locus phylogenetic assessment of North Amer-ican red squirrels and test three alternative hypotheses based onprevious studies (Fig. S1). In the first hypothesis (H1 – Fig. 1) we

apply the current taxonomy (Thorington and Hoffmann, 2005) ofthree species (T. douglasii, T. hudsonicus, and T. mearnsi). Consider-ing that the specific status of T. mearnsi is tenuous (Arbogast et al.,2001), we test a second hypothesis (H2 – Fig. 2) proposed byArbogast et al. (2001) that also posits three differently circum-scribed species: T. douglasii (inclusive of T. mearnsi), T. fremonti,and T. hudsonicus (includes all currently recognized lineages exceptputative T. fremonti). Finally, we consider a third hypothesis (H3 –Fig. 3) treating nine well-supported mtDNA lineages as terminaltaxa, all documented across independent studies (Arbogast et al.,2001; Barber, 2007; Chavez et al., 2014; this study). These threehypotheses require few assumptions and reflect realistic alterna-tive relationships based on previous literature. Following genetic-based species delimitation, we incorporate niche modeling toassess geographic stability through time and ecological divergenceamong taxa. Finally, we use results from both multiple loci and cli-matic data to explore evolutionary complexity across North Amer-ica, demonstrating different modes of speciation within thiscomplex that are reflected more broadly among mammals. Criti-cally, for forest ecosystemmanagement, we identify areas that sus-tain key dynamic evolutionary processes as well as isolatedpopulations needing further evaluation.

2. Methods

2.1. Study area, sampling and sequencing

We obtained specimens from museum collections that maxi-mize geographic sampling through the known range of red squir-rels (Figs. 1 and 3), including samples from all species andlocalities coincident with the geographic distribution of 22 of 28subspecies (Hall, 1981). We used 252 specimens (Appendix A)including 40 T. douglasii, 209 T. hudsonicus, and 3 T. mearnsi from109 general collecting localities (Fig. 3; 153 specific localities;Appendix A). Sequences for 123 individuals for at least 1 gene wereretrieved from GenBank. We then sequenced 129 more specimensfor 1–4 independent loci where a subset of 60 samples (including34 specimens from Chavez et al., 2014) were represented by bothmtDNA and nuDNA. In total we obtained 3192 bp of sequence dataacross all loci. Datasets included 771 base pairs (bp) of mtDNAcytochrome-b gene (Cytb; n = 252), 449 bp of the glutamate decar-boxylase 2 gene (intron 1; GAD2; n = 60), 666 bp of theganglioside-induced differentiation-associated protein 1 gene(intron 1; GDAP1; n = 58), 630 bp of the UPF0351 proteinC9orf32 gene (intron 1; METTL11A; n = 59), and 676 bp of the pro-lyl 4-hydroxylase alpha-2 subunit precursor gene (intron 1;P4HA2; n = 60).

We extracted genomic DNA from tissue (stored at �80 �C orethanol preserved) through standard salt extraction, followed byPCR amplification and cycle sequencing with ABI BigDye v3.1(Applied Biosystems, Foster City, California) following protocolsof Platt et al. (2007). Primers for Cytb were MVZ05/MVZ16(Smith and Patton, 1993). Other primers are listed in Chavezet al. (2014). We conducted automated sequencing of complimen-tary strands using the Applied Biosystems 3110 DNA sequencer atthe University of New Mexico.

We edited and aligned sequences in Sequencher v4.8 (Gene-codes, Ann Arbor, Michigan) and verified visually. We translatedprotein-coding sequences for Cytb to amino acids and examinedthem for internal stop codons, rates of transition/transversionchanges, and relative 1st, 2nd, and 3rd position changes in codonsthat might indicate a Numt. None was found. We compared com-plementary strands of DNA and deposited contiguous sequencesin GenBank (Appendix A). We inferred alleles of heterozygotesusing Phase (Stephens et al., 2001), by accepting results with a

https://www.researchgate.net/publication/230218003_Comparative_phylogeography_as_an_integrative_approach_to_historical_biogeography?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/6668157_Cryptic_species_as_a_window_on_diversity_and_conservation?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/12675238_Channell_R_Lomolino_MV_Dynamic_biogeography_and_conservation_of_endangered_species_Nature_403_84-86?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/222669908_A_phylogeographic_perspective_on_endemism_in_the_Alexander_Archipelago_of_the_North_Pacific?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/222520150_Conservation_of_highly_fragmented_systems_The_north_temperate_Alexander_Archipelago?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/271677269_Geographic_Variation_in_New_Mexican_Chickarees?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/26725902_When_Cold_is_Better_Climate-Driven_Elevation_Shifts_Yield_Complex_Patterns_of_Diversification_and_Demography_in_an_Alpine_Specialist_American_Pika_Ochotona_princeps?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/228060748_Northern_Host-Parasite_Assemblages_History_and_Biogeography_on_the_Borderlands_of_Episodic_Climate_and_Environmental_Transition?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/265462716_Multilocus_phylogeography_and_systematic_revision_of_North_American_water_shrews_genus_Sorex?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/null?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/257646459_Using_biogeographical_history_to_inform_conservation_The_case_of_Preble's_meadow_jumping_mouse?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/222320481_The_diversification_of_South_American_murid_rodents_Evidence_from_mitochondrial_DNA_sequence_data_for_the_akodontine_tribe?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230233677_Molecular_Systematics_and_the_Conservation_of_Rare_Species?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/12078983_A_New_Statistical_Method_for_Haplotype_Reconstruction_from_Population_Data?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/261997310_Divergence_with_gene_flow_within_the_recent_chipmunk_radiation_Tamias?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/27397977_The_Mammals_of_North_America_Vol_1?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/null?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/249646273_How_to_fail_at_species_delimitation?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/249646273_How_to_fail_at_species_delimitation?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

Hypothesis H1

1.0

1.0

Elevation

Species

Fig. 1. Distribution and current taxonomy of North American tree squirrels (Tamiasciurus). Data points indicate localities of all samples considered for niche analyses. Insetprovides the results of a ⁄BEAST species tree analysis (including empirical topology and nodal support values) considering systematic hypothesis H1 with three species as T.hudsonicus, T. douglasii, and T. mearnsi. Downloaded from Wikipedia March 2015. Author: Cephas. This file is licensed under the Creative Commons Attribution-Share Alike3.0. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License Version 1.2 or any later version publishedby the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled GNU FreeDocumentation License.

172 A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182

probability >90%, using a burn-in of 10,000 iterations and a runlength of 10,000 iterations (McCormack et al., 2011). Among 5runs, we retained results from the run with best goodness-of-fitto an approximate coalescent model, resulting in 2 nuclear haplo-type sequences or alleles per individual, using only 1 allele perindividual (chosen randomly) for analyses.

2.2. Multi-locus gene trees and evolutionary rates

We estimated independent genealogies and haplotype net-works for all loci. We determined DNA substitution models foreach locus using MrModeltest v2.3 (Nylander, 2004) under theAkaike Information Criterion. We performed Markov chain MonteCarlo (MCMC) searches in MrBayes v3.1 (Ronquist andHuelsenbeck, 2003), incorporating the short-branch method(Marshall, 2010). Stationarity of MCMC runs was assessed in Tracerv1.5.0 (Rambaut and Drummond, 2007). Phylograms were mid-point rooted and visualized with posterior probabilities >0.75included in FigTree v1.3.1 (Rambaut, 2009). For all loci, weassigned haplotypes using DnaSP v5 (Librado and Rozas, 2009).Median-joining haplotype networks for all loci were constructedin Network v4.610 (Bandelt et al., 1999). To assess the timing ofcladogenesis and demographic changes among lineages, we calcu-lated average and lineage-specific evolutionary substitution ratesutilizing a time to coalescence of the entire genus as estimatedby Chavez et al. (2014). In BEAST v1.6.1 (Drummond andRambaut, 2007), we produced a chronogram based on the Cytblocus by fixing the time to most recent common ancestor (TMRCA)

for all samples at 0.4229 Ma (normal distribution, standard devia-tion 0.1; Chavez et al., 2014), applying a relaxed clock: uncorre-lated log-normal, and set the substitution rate to be estimated.We used the HKY + G model of substitution, empirical base fre-quencies, partitioned into three codon positions, and a constantpopulation size tree prior. Three separate runs of 100 million iter-ations were run, sampling trees to file every 1000 iterations. Toretrieve mean lineage-specific rates for each nominal lineagerecovered from the Cytb chronogram, we weighted all branchspecific rates by branch length (Hope et al., 2014b) according tothe equation

P(rate � length)/P (length).

2.3. Mitochondrial diversity and demographics

Genetic diversity analyses for Cytb used all available red squir-rel sequences, performing analyses separately for major lineageassignments (Table 1). In DnaSP, for each group we assessedgenetic diversity including number of segregating sites (S), haplo-types (h), haplotype diversity (Hd), nucleotide diversity (p), andtested for demographic expansion by calculating Tajima’s D(Tajima, 1989) and R2 (Ramos-Onsins and Rozas, 2002).

To calculate population size change through time, we imple-mented the coalescent Bayesian Skyline Plot (BSP) tree prior inBEAST. Results of Bayes factor tests to determine the best clockmodel indicated no significant difference between strict andrelaxed clocks for any lineage, so we utilized a strict clock, andran MCMC analyses (length of chain = 100 M logging trees every

https://www.researchgate.net/publication/12967964_Median-Joining_Networks_for_Inferring_Intraspecific_Phylogenies?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/null?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/null?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/263095830_Accounting_for_rate_variation_among_lineages_in_comparative_demographic_analyses?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/24257674_DnaSP_v5_A_Software_for_Comprehensive_Analysis_of_DNA_Polymorphism_Data?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/44651600_Cryptic_Failure_of_Partitioned_Bayesian_Phylogenetic_Analyses_Lost_in_the_Land_of_Long_Trees?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/45493340_Calibrating_divergence_times_on_species_trees_versus_gene_trees_Implications_for_speciation_history_of_Aphelocoma_jays?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/285805344_MrModeltest_V2_Program_Distributed_by_the_Author?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/20643397_Statistical_method_for_testing_the_neutral_mutation_hypothesis_by_DNA_Polymorphism?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/11021975_Statistical_Properties_of_New_Neutrality_Tests_Against_Population_Growth?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/10618138_MRBAYES_3_Bayesian_Phylogenetic_inference_under_mixed_models?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/10618138_MRBAYES_3_Bayesian_Phylogenetic_inference_under_mixed_models?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

-80° W-100° W-120° W

-120° W-140° W-160° W-180°

50° N

5 ° N

40° N

40° N

30° N

30° N

20° N

Hypothesis H2

1.0

.47

0 500 1,000250

Kilometers

SpeciesT. douglasiiT. hudsonicusT. fremontiT. douglasii +T. hudsonicusT. fremonti +T. hudsonicus

0

10

20

30

30 40

50

~ 80

90

100

Niche IdenititySchoener’s D

Fig. 2. Bioclimatic envelope models for North American red squirrels considering three species according to systematic hypothesis H2 as T. douglasii (green), T. fremonti(brown), and T. hudsonicus (dark blue). We used model selection and applied a minimum of 90% of training samples as our threshold for each species. Potential geographicoverlap between species is depicted with intermediate colors (cyan between T. hudsonicus and T. douglasii, and gray between T. hudsonicus and T. fremonti). Left inset providesthe results of ⁄BEAST species tree analysis considering H2, including empirical topology and nodal support values. Top inset shows results of niche identity test, where allpairwise comparisons indicate significant ecological differentiation between species. The histogram provides expected niche overlap when pairwise species are pooled andrandomized, and arrows indicate empirical niche overlap between species (Schoener’s D). (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182 173

1000), assigning lineage-specific rates calculated from the Cytbchronogram.

2.4. Systematic scenarios

Using a model comparison technique that applies Bayes Factors(2lnBf) or Akaike information criterion through Markov chainMonte Carlo analysis (AICM) to test between alternative scenariosof species delimitation (Aydin et al., 2014; Grummer et al., 2014)allows flexibility for alternative hypotheses and is robust to sys-tems where historical introgression is suspected (Grummer et al.,2014). We performed model comparison tests of alternative sys-tematic hypotheses among designated lineages of North Americanred squirrels considering all loci within a species-tree frameworkimplemented using ⁄BEAST (Heled and Drummond, 2010). Because⁄BEAST accommodates different numbers of gene copies for eachtaxon, we included all samples per lineage that were sequencedfor more than a single locus.

We performed Bayes Factor and AICM pairwise tests of alterna-tive hypotheses (Table 2; Fig. S1). All hypotheses considered sam-ples (n = 60) representing various combinations of the nine well-supported monophyletic mtDNA (Cytb) lineages: Two widespreadlineages (Continental and Northwestern), a Western lineage (com-bining T. douglasii and Baja California samples unless otherwisenoted), Vancouver Island, Central Rockies, North Pacific Coast,Sacramento Mountains (NM), Southern Rockies, and SouthwesternSky Islands (Fig. 1). Despite a high likelihood of ongoing gene flow

within and among certain lineages, all were included to assess rel-ative nodal support for clade relationships considering multipleloci under three different hypotheses. Hypothesis H1 consideredthe currently assigned taxonomy (T. douglasii, T. hudsonicus, andT. mearnsi; Fig. 2). Hypothesis H2 considered three species hypoth-esized by Arbogast et al. (2001): T. douglasii (including T. mearnsi),T. fremonti (Southern Rockies, Sacramento Mountains, and South-western Sky Island), and T. hudsonicus (Continental, Northwestern,Southeast Alaska, Vancouver Island and Central Rockies). Hypoth-esis H3 considered all Cytb lineages as independent ‘‘species”. Allhypotheses were compared statistically using both all loci and onlynuclear loci.

We produced a separate traits file indicating lineage assign-ments for all samples under each hypothesis. In BEAUti, for eachhypothesis, we unlinked datasets for each locus across all parti-tions, assigned substitution models by locus, and applied empiricalbase frequencies. We applied a relaxed clock: uncorrelated log-normal to Cytb as the ucld standard deviation of the rate for thisgene when analyzed independently significantly deviated fromzero, but assigned a strict clock to nuclear loci to reflect lowerdivergence over the evolutionary timeframe investigated. Wechose a Yule tree prior with piecewise linear and constant rootand assigned proper ploidy to each locus. Two independent runsfor each hypothesis used MCMC chains of 100 million generationssampling trees every 10,000.

In Tracer, we ran model comparisons to identify the hypothesisthat best explained the data considering marginal likelihood

https://www.researchgate.net/publication/265597865_Marginal_Likelihood_Estimate_Comparisons_to_Obtain_Optimal_Species_Delimitations_in_Silene_sect_Cryptoneurae_Caryophyllaceae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/258827498_Species_Delimitation_Using_Bayes_Factors_Simulations_and_Application_to_the_Sceloporus_scalaris_Species_Group_Squamata_Phrynosomatidae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/258827498_Species_Delimitation_Using_Bayes_Factors_Simulations_and_Application_to_the_Sceloporus_scalaris_Species_Group_Squamata_Phrynosomatidae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/258827498_Species_Delimitation_Using_Bayes_Factors_Simulations_and_Application_to_the_Sceloporus_scalaris_Species_Group_Squamata_Phrynosomatidae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/38083310_Bayesian_Inference_of_Species_Trees_from_Multilocus_Data?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

Fig. 3. Left – Chronogram of evolutionary relationships among North American red squirrels (Tamiasciurus) based on 771 bp of the mtDNA Cytochrome b gene with a rootcoalescence estimate from Chavez et al. (2014). Phylogeny estimation was performed in BEAST, providing both posterior probability nodal support values (left of node) andcoalescence times (millions of years; italics, right of node). The topology is presented as ultrametric and proportional to an evolutionary timeline (bottom) that extends fromthe present (right) to the past (left). Tip labels indicate general collecting localities (Appendix A). Major lineages are color coded and coincident with geography. Tip labelsindicate sampling localities. Right – Map of the study area indicating general specimen localities for samples of North American red squirrels included in the present study.Locality numbers correspond to those listed in Appendix A. Left inset provides finer detail of localities along the North Pacific Coast. Bottom-right inset provides finer detail oflocalities within the Southwestern region. Localities are colored according to major mtDNA lineages. Circles with multiple colors indicate general localities where multiplemtDNA lineages are sympatric. For insets, insular populations are labeled by island or mountain range. Top inset provides the results of a ⁄BEAST species tree analysis(including empirical topology and nodal support values) considering systematic hypothesis H3 with nine independent taxa color-coded according to mtDNA lineages. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1Genetic polymorphism and demographic statistics considering 771 bp of the mtDNA Cytochrome b gene among major North American red squirrel lineages. Abbreviationsinclude n = sample size; S = number of segregating sites; h = number of haplotypes; Hd = haplotype diversity; p = nucleotide diversity; D = Tajima’s D (with associated P-value);R2 = Ramos-Onsins and Rozas R2 (with associated P-value); l = lineage-specific substitution rate (percent per million years) considering rates weighted by branch lengths foreach lineage from the Cytochrome b BEAST chronogram. The rate given for all lineages combined was calculated in BEAST using a root calibration of 0.4229 Myr (HPD: 0.2979–0.5538; Chavez et al., 2014). N/A = not applicable. Bold values are statistically significant at P < 0.05.

n S h Hd p D P(D) R2 P(R2) l

All 252 134 142 0.982 0.0181 N/A N/A N/A N/A 5.23Central Rockies 11 4 4 0.600 0.0015 �0.542 0.324 0.142 0.107 5.14Continental 60 41 34 0.866 0.0045 �2.167 0.001 0.046 0.000 5.20Northwestern 52 45 32 0.924 0.0031 �2.593 0.000 0.022 0.000 5.35North Pacific Coast 20 15 16 0.963 0.0040 �1.192 0.132 0.088 0.056 5.03Sacramento Mountains 10 11 7 0.933 0.0035 �1.407 0.096 0.124 0.056 5.26Southern Rockies 15 14 10 0.924 0.0035 �1.509 0.064 0.079 0.000 5.42Southwestern Sky Islands 22 17 12 0.905 0.0037 �1.413 0.062 0.073 0.011 5.26Vancouver Island 19 8 7 0.544 0.0012 �2.022 0.001 0.084 0.002 5.21Western 42 35 26 0.962 0.0067 �1.325 0.074 0.065 0.059 5.25

174 A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182

estimates of competing hypotheses using both 2lnBf and 2AICMscores. Scores were ranked for each hypothesis only with respectto the ‘‘best” hypothesis (that with the marginal likelihood esti-mate closest to zero). As described in Grummer et al. (2014), andbased on the recommendations of Kass and Raftery (1995), thestrength of support of 2lnBf for the best hypothesis (labeled as N/A) compared with competing hypotheses was as follows: 0–2means no significant support; 2–6 means positive support for thebest hypothesis (N/A) over this competing scenario; 6–10 means

strong support for the best hypothesis; and >10 means decisiverejection of the competing hypothesis compared with the besthypothesis. As in Grummer et al. (2014), we recognized distinctlineages from hypotheses (topologies) where 2lnBf support forthe leading topology was >10 in pairwise comparisons with com-peting hypotheses. Similarly for AICM which equates using twotimes the sample mean log likelihood, we use a cutoff of >20 for2AICM (multiplied by 2 for consistency with Bayes Factormethods).

https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/258827498_Species_Delimitation_Using_Bayes_Factors_Simulations_and_Application_to_the_Sceloporus_scalaris_Species_Group_Squamata_Phrynosomatidae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/258827498_Species_Delimitation_Using_Bayes_Factors_Simulations_and_Application_to_the_Sceloporus_scalaris_Species_Group_Squamata_Phrynosomatidae?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

Table 2AICM and Bayes Factor (2lnBf) results considering marginal likelihood estimates from alternative species-tree hypotheses of relationships among major lineages of NorthAmerican red squirrels (see Fig. S1 for scenario topologies). The hypothesis receiving the best marginal-likelihood score is indicated by N/A. 2AICM and 2lnBf scores for otherscenarios reflect comparative strength of the best supported hypothesis, where for 2lnBf a score of 0–2 means ‘‘not worth more than a bare mention”, 2–6 means ‘‘positive”support for the best hypothesis over that alternative, 6–10 provides ‘‘strong” support, and >10 means ‘‘decisive” support in distinguishing between best and competinghypotheses (after Kass and Raftery, 1995). For 2AICM scores, we consider >20 as decisive, 12–20 as strong, and 4–12 as positive support for the best hypothesis over thatalternative. Hypotheses were analyzed considering all five loci (both mtDNA and nuDNA) and using only the four nuclear loci.

Scenario Description All five loci Four nuDNA loci

2AICM 2lnBf 2AICM 2lnBf

H1 3 species as currently described (T. douglasii, T. hudsonicus, T. mearnsi) N/A 10.28 5.72 10.74H2 3 species as per Arbogast et al. (2001; T. douglasii, T. hudsonicus, T. fremonti) 27.85 N/A N/A N/AH3 All lineages considered separately 104.10 22.76 31.70 16.06

A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182 175

2.5. Bioclimatic envelope modeling

We used bioclimatic envelope modeling (BEM, Araújo andPeterson, 2012) to characterize occupied distributions and recon-struct paleodistributions of each species. We used these niche-based tests as an independent perspective of ecological divergenceamong putative species using a three-step approach: (1) recon-struct contemporary distributions and project (transfer) the finalmodel to ancestral climate conditions (i.e., mid-Holocene andLGM); (2) assess spatial overlap of predicted distributions for allperiods; and (3) conduct niche identity tests using geographic-based approaches (Warren et al., 2008, 2010).

2.5.1. Occurrence and environmental dataBEMs are often constructed from two forms of data: occurrence

records and environmental layers. We downloaded all availablerecords of North American red squirrels from VertNet (accessedFebruary 2015) and conducted a series of screening and filteringsteps to reduce bias. We removed all samples with a georeferencederror >5 km, those that failed to provide an estimate of error, andthose that were outside the known distribution of Tamiasciurus.We also removed samples prior to 1940. To reduce spatial autocor-relation, we used a custom aggregated reduction technique(Fourcade et al., 2014; Warren et al., 2014) by retaining a singlerandom sample within 10 km2 radius and discarding other spa-tially redundant locations.

To assess lineage-based distributional overlap and test forrecent niche divergence among evolutionarily independent groupswe used mtDNA haplotypes to quantify areas occupied by lineages.First, we added a minimum convex polygon in ArcGIS v10.2 to eachlineage-based set of points and applied a 200 km buffer to assignall points within each polygon to respective mtDNA lineages. Wethen reassigned mtDNA samples to putative species identifiedfrom the best model in tests of evolutionary hypotheses.

The geographic extent of species is an important considerationwhen using correlative models (Barve et al., 2011). However, thetotal extent of Tamiasciurus is likely too broad considering ourgoals of pairwise tests between geographically limited species.Based on the geographic distribution of putative species, wedivided the study area using the three regional clades (systematichypothesis H2; Table 1; Fig. S1) and delimited their extents to100 km beyond the N/S and E/W extremes of each distribution.

Next, we generated bias files for each species to optimize back-ground and occurrence point selection within study extents. Biasfiles are frequently used in correlative modeling to avoid oversam-pling and can offset effects of geographic biases often associatedwith coordinate-based data, decreasing commission (false-positive) error rates (Anderson and Raza, 2010; Barve et al.,2011; Merow et al., 2013). We used SDMToolBox v1.1c (Brown,2014) to generate the bias files for each species, using the ‘‘Sampleby Buffered Local Adaptive Convex-Hull” tool that limits back-ground points (Thuiller et al., 2009; Barbet-Massin et al., 2012)and set the buffer distance to 75 km and alpha parameter to four.

We used 19 bioclimatic (temperature and precipitation) vari-ables from the WorldClim database to quantify ecological toler-ances of each lineage. These landscape-level data are often usefulfor assessing both current and paleodistributions (Malaney andCook, 2013; Waltari and Guralnick, 2009; Waltari et al., 2007),and also for assessing Grinnellian-based niche divergence(McCormack et al., 2010). Then, we identified tightly correlated(P0.80) and removed redundant variables, favoring those moretemporally inclusive (e.g., bio11 – mean temperature of the coldestquarter versus bio6 – minimum temperature of the coldest month)or relevant to squirrel ecology (e.g., cone mast is often linked toseasonal conditions; bio15 – precipitation seasonality).

2.5.2. Correlative modelingWeusedMaxEnt version 3.3.3e (Phillips and Dudik, 2008) due to

superior performance over other correlative modeling approaches(Elith et al., 2006; Peterson et al., 2007) and ability to provide statis-tical comparisons among models (Elith et al., 2011; Merow et al.,2013). MaxEnt uses presence data in comparison with randombackground samples to estimate species distributions (Elith et al.,2011; Guillera-Arroita et al., 2015; Merow et al., 2013), but is opti-mized with multiple default settings that require testing prior tomodeling to enhance model performance (Anderson andGonzalez, 2011). Consequently, we conducted criterion-basedmodel selection (Burnham and Anderson, 2002; Warren andSeifert, 2011) by assessing different combinations of the featureclass types (FC) and regularization multipliers (RM) usingENMTools v.1.4.4 (Warren et al., 2010).We constructed singlemod-els for each species by applying alternate FCs (e.g., L – linear; LQ –linear and quadratic; H – hinge; LQH – linear, quadratic, and hinge;LQHPT – LQH, product, and threshold) and alternate RMs rangingfrom 0.01 to 5.0 at 0.5 intervals. In sum, we tested 55 alternatemodels for each species (5 FCs � 11 RMs), identifying optimummodel settings using corrected Akaike information criterion (AICc).

We then conducted final modeling for each species using 20replicates, log-scale outputs, and set number of iterations to 5 k,keeping the default convergence threshold (=0.00001). Duringreplicate models, we applied the bias file for background sampling,retained 20% of presence localities as a training dataset, and usedsubsampling run type (jackknife approach) because of small sam-ple sizes for some mtDNA lineages (Pearson et al., 2007;Shcheglovitova and Anderson, 2013). Standard deviations acrossall replicates were minor.

MaxEnt produces a set of continuous surfaces and we used themean of replicated models to represent relative suitability/likeli-hood (Guillera-Arroita et al., 2015; Richmond et al., 2010). Toassess differences between high suitability for each species, weconstructed a single raster by calculating the difference betweenthe continuous mean logistic outputs of comparative lineages (Dif-fAB = species A – species B). Further, we created a binary distribu-tion of suitable and unsuitable areas using the equal trainingsensitivity and specificity logistic threshold for the spatial projec-tion of final models (Pearson et al., 2007) and used the

https://www.researchgate.net/publication/233382686_Species-specific_tuning_increases_robustness_to_sampling_bias_in_models_of_species_distributions_an_implementation_with_MAXENT?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/233382686_Species-specific_tuning_increases_robustness_to_sampling_bias_in_models_of_species_distributions_an_implementation_with_MAXENT?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227738980_The_effect_of_the_extent_of_the_study_region_on_GIS_models_of_species_geographic_distributions_and_estimates_of_niche_evolution_Preliminary_tests_with_montane_rodents_genus_Nephelomys_in_Venezuela?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230691635_The_crucial_role_of_the_accessible_area_in_ecological_niche_modeling_and_species_distribution_modeling?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230691635_The_crucial_role_of_the_accessible_area_in_ecological_niche_modeling_and_species_distribution_modeling?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230691635_The_crucial_role_of_the_accessible_area_in_ecological_niche_modeling_and_species_distribution_modeling?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/261769474_SDMtoolbox_A_python-based_GIS_toolkit_for_landscape_genetic_biogeographic_and_species_distribution_model_analyses?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/261769474_SDMtoolbox_A_python-based_GIS_toolkit_for_landscape_genetic_biogeographic_and_species_distribution_model_analyses?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/256476402_Model_Selection_and_Multimodel_Inference_A_Practical_Information-Theoretic_Approach?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227546243_Novel_methods_improve_prediction_of_species'_distributions_from_occurence_data?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227674246_A_statistical_explanation_of_MAXENT_for_ecologists?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227674246_A_statistical_explanation_of_MAXENT_for_ecologists?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227674246_A_statistical_explanation_of_MAXENT_for_ecologists?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/262267791_Mapping_Species_Distributions_with_MAXENT_Using_a_Geographically_Biased_Sample_of_Presence_Data_A_Performance_Assessment_of_Methods_for_Correcting_Sampling_Bias?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/257646459_Using_biogeographical_history_to_inform_conservation_The_case_of_Preble's_meadow_jumping_mouse?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/257646459_Using_biogeographical_history_to_inform_conservation_The_case_of_Preble's_meadow_jumping_mouse?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/38098199_Does_niche_divergence_accompany_allopatric_divergence_in_Aphelocoma_jays_as_predicted_under_ecological_speciation_Insights_from_tests_with_niche_models?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/264532217_A_practical_guide_to_MaxEnt_for_modeling_species'_distributions_What_it_does_and_why_inputs_and_settings_matter?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/264532217_A_practical_guide_to_MaxEnt_for_modeling_species'_distributions_What_it_does_and_why_inputs_and_settings_matter?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/264532217_A_practical_guide_to_MaxEnt_for_modeling_species'_distributions_What_it_does_and_why_inputs_and_settings_matter?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/264532217_A_practical_guide_to_MaxEnt_for_modeling_species'_distributions_What_it_does_and_why_inputs_and_settings_matter?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227714587_ORIGINAL_ARTICLE_Predicting_species_distributions_from_small_numbers_of_occurrence_records_a_test_case_using_cryptic_geckos_in_Madagascar?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227714587_ORIGINAL_ARTICLE_Predicting_species_distributions_from_small_numbers_of_occurrence_records_a_test_case_using_cryptic_geckos_in_Madagascar?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227524719_Transferability_and_model_evaluation_in_ecological_niche_modeling_A_comparison_of_GARP_and_Maxent?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227635331_Modeling_of_species_distributions_with_MAXENT_new_extensions_and_a_comprehensive_evaluation?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259217968_Estimating_optimal_complexity_for_ecological_niche_models_A_jackknife_approach_for_species_with_small_sample_sizes?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227733897_BIOMOD_-_A_platform_for_ensemble_forecasting_of_species_distributions?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/229544335_Ecological_niche_modelling_of_montane_mammals_in_the_Great_Basin_North_America_Examining_past_and_present_connectivity_of_species_across_basins_and_ranges?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/6216408_Locating_Pleistocene_Refugia_Comparing_Phylogeographic_and_Ecological_Niche_Model_Predictions?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/250077100_Ecological_niche_modeling_in_Maxent_The_importance_of_model_complexity_and_the_performance_of_model_selection_criteria?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/250077100_Ecological_niche_modeling_in_Maxent_The_importance_of_model_complexity_and_the_performance_of_model_selection_criteria?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/299269998_Environmental_niche_equivalency_versus_conservatism_quantitative_approaches_to_niche_evolution_vol_62_pg_2868_2008?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227710797_ENMTools_A_toolbox_for_comparative_studies_of_environmental_niche_models?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/227710797_ENMTools_A_toolbox_for_comparative_studies_of_environmental_niche_models?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259548076_Incorporating_model_complexity_and_spatial_sampling_bias_into_ecological_niche_models_of_climate_change_risks_faced_by_90_California_vertebrate_species_of_concern?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/229149956_Selecting_Pseudo-Absences_for_Species_Distribution_Models_How_Where_and_How_Many?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/46579656_Is_the_Climate_Right_for_Pleistocene_Rewilding_Using_Species_Distribution_Models_to_Extrapolate_Climatic_Suitability_for_Mammals_across_Continents?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/270598364_Is_my_species_distribution_model_fit_for_purpose_Matching_data_and_models_to_applications?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/270598364_Is_my_species_distribution_model_fit_for_purpose_Matching_data_and_models_to_applications?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

176 A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182

threshold-applied distributions to characterize paleodistributionsand assess distributional shifts (see below).

2.5.3. PaleodistributionsWe reconstructed paleodistributions of individual lineages to

understand the history of geographic distributions, frequentlyyielding insights of ancestral areas as well as regions of historicadmixture (Pelletier et al., 2015). Specifically, we projected predic-tions for the mid-Holocene (contemporary models; i.e., YoungerDryas) and late-Pleistocene (i.e., Last Glacial Maximum – LGM).Considering many North American mammals have shifted distri-butions (Malaney and Cook, 2013; Lessa et al., 2003), we expandedthe spatial extents through past timeframes by 500 km to includeareas nearby that might have been historically occupied by ances-tral populations. Further, we validated paleodistribution recon-structions with fossil records obtained from FaunMap database(accessed March, 2015). We then assessed spatial overlap amongclades using techniques analogous to contemporary overlapmeasurements.

2.5.4. Model-based niche divergenceTo assess the degree of niche overlap among the major clades,

we used the G-space niche-identity test using the D statistic(Schoener, 1968; Warren et al., 2008), which ranges from 0 (discor-dant) to 1.0 (identical). The niche-identity test is used to assesswhether measured overlap is significantly different (one-tailedtest) from an underlying null expectation. This approach random-ized samples from each lineage pair using 100 pseudoreplicates foreach pairwise test.

3. Results

3.1. Gene trees and evolutionary rates

A previously calculated age for the TMRCA of all North Americanred squirrels of 0.4229 Myr (Chavez et al., 2014) provided an aver-age mutation rate of 5.23% per million years (10.46% divergencerate) for the Cytb dataset (Table 1). Substitution rates ranging from5.03% to 5.46% per million years were recovered for nominal lin-eages within the red squirrel Cytb genealogy (Table 1). The chrono-gram indicates nine well-supported, spatially-structured lineages(Fig. 3). Two lineages exhibit broad distributions, often within pre-viously glaciated areas, indicative of post-glacial population expan-sion. The Continental lineage is broadly distributed across NorthAmerica whereas the Northwestern lineage is coincident with theRocky Mountain cordillera. These two lineages together form awell-supported clade distinct from all other North American redsquirrels. Other well-supported lineages include the Central Rock-ies with a limited distribution within the inland forests of Idaho,western Montana, and eastern Oregon. The North Pacific Coast(NPC) lineage is only distributed on islands of the Alexander Archi-pelago (Betton, Grant, Gravina, Kuiu, Mitkof, Revillagigedo, andZarembo islands), and in British Columbia on Calvert and Swindleislands (Fig. 3). Mitochondrial signatures indicative of both NPCand Northwestern lineages occur on Gravina and Revillagigedoislands of the southern Alexander Archipelago. The Northwesternlineage also occurs on Admiralty (introduced), Tatoosh, and Wran-gell islands. The Western mtDNA lineage is paraphyletic withrespect to Vancouver Island (due to recognized mito-nuclear dis-cord) and is distributed from Baja California (T. mearnsi) north tosouthern British Columbia. Finally, substructure within the well-supported Southwestern clade includes lineages associated withthe Sacramento Mountains of New Mexico, the Southern Rockiesof Colorado and northern New Mexico, and two paraphyletic

lineages associated with the Southwestern Sky Island archipelago(Fig. 3).

The initial dichotomy within the Cytb chronogram is dated to�400 ka. Continental and Northwestern lineages were estimatedto have split at 200 ka. Coalescence of all Southwestern lineagesis coincident with the Illinoian glacial period (�160 ka) althoughinitial divergence from other squirrel lineages is dated >250 ka.Coalescence of most terminal lineages is

A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182 177

and Southwestern Sky Islands lineages, and the American redsquirrel, T. hudsonicus consisting of the remaining 5 lineages.

3.4. Bioclimatic envelope modeling

3.4.1. Niche variables and occurrence pointsWe obtained 20 niche variables that exhibited varying levels of

importance and correlation across the study region (Table S1). Foroccurrence points, we obtained a total of 1160 T. douglasii (includ-ing T. mearnsi), and 3413 T. hudsonicus records. After processing(screening, partitioning, thinning, and clipping), we identified279 T. douglasii and 740 T. hudsonicus records for reconstructingdistributions and conducting niche tests. We then assigned 59 ofthe T. hudsonicus records from the Southwest as T. fremonti usinggenetic data as a guide.

3.4.2. Model performanceWe found that optimized settings reflected a best-fit for our

study system using model selection that increased the regulariza-tion penalty, simplified features, and reduced the overall dimen-sionality of final models. MaxEnt performed well with meantraining AUC >0.85 for all analyses, often a positive indication ofoptimized models. Randomly withheld test data were predictedsignificantly better than random with AUC >0.90 for all analyses(Table 3) and AUC standard deviations all 73% of thevariation was accounted for by winter precipitation(bio19 = 39.7%), precipitation seasonality (bio15 = 20.0%), andisothermality (bio3 = 13.5%). For T. hudsonicus, over 50% variableimportance was due to isothermality (31.3%) and winter precipita-tion (21.5%), with mean annual temperature (bio1 = 12.7%) thethird most important, totaling 79.3%. The three most importantvariables (65.5%) for T. fremonti included average summer temper-ature (bio5 = 50.0%), temperature seasonality (bio4 = 13.5%), andlike other species, winter precipitation (15.8%).

Current distributions predicted for each species overlap (Fig. 2),with T. hudsonicus sharing predicted range with T. douglasii in thePacific Northwest (overlap = 77,754 km2) and with T. fremonti inthe Southern Rocky Mountains (overlap = 71,408 km2).

3.4.4. PaleodistributionsWe document distributional change for all species between the

LGM, mid-Holocene, and present. Even after applying thresholds of

Table 3Relative importance of niche model variables. During modeling, we conductedparameters from model selection. The training, test, and standard deviation of AUAUC metrics are the relative contributions of variables. Variables shown contcorresponding to the top three variables in each model. Symbols reflect results of tin isolation (⁄), the variable that decreased the gain most when omitted (#), and

AUC/variable T. douglasii

Test AUC 0.942AUC SD 0.008Annual Temperature (Bio1)Isothermality (Bio3) 13.5Temperature Seasonality (Bio4)Summer Temperature (Bio5)Precipitation Seasonality (Bio15) 20.0#,$

Winter Precipitation (Bio19) 39.7⁄

currently occupied distributions and projecting to historical condi-tions (i.e., mid-Holocene and LGM), most fossil records (>80%,Figs. S4–S6) were accurately predicted. Predicted ancestral areas(LGM) for T. douglasii were centered on northern California(Fig. S4) with isolated regions in southern California and northernBaja, and although generally further south, total predicted area wasonly slightly larger than mid-Holocene and today. Predicted distri-butions shifted to higher elevations and more northern latitudesduring the Holocene and are most restricted in total range at pre-sent. For T. fremonti, LGM distributions were projected to be >2�the extent of present distributions, with a retraction to higher ele-vations during the Holocene rather than a shift across latitude(Fig. S5). For T. hudsonicus, latitudinal shifts appear to have beenthe dominant change from LGM to mid-Holocene and today(Fig. S6). Potential LGM occupation included ancestral regionsalong the North Pacific Coast and within eastern Beringia (Alaska).

3.4.5. Stable regions for squirrelsBy overlaying predictions from all models, we identified stable

regions for all three species, although eco-geographic stability var-ied. For T. douglasii, three regions have remained relatively stablebetween the LGM and today, including medium elevations in theSierra Nevada of eastern California, the Coast Ranges in northwest-ern California and Oregon, and high elevations of southern Califor-nia and Baja Mexico. The total combined stable area for T. douglasiiis ca. 69,000 km2 (20.8% of the present distribution). For T. fremonti,we detected three temporally stable areas largely coincident inextent with the current range, including the Rocky Mountains ofColorado and New Mexico, the Sacramento Mountains in south-eastern New Mexico, and the Mogollon Mountains of New Mexicoand Arizona. When combined, these stable areas for T. fremontitotaled ca. 111,000 km2 (62.9% of the present distribution). In con-trast to T. douglasii and T fremonti, T. hudsonicus had few consis-tently occupied regions between the LGM and today totaling10% to each model and were used for ecological interpretation,he jackknife test of variable importance showing the highest gain when usedthe variable with the highest permutation importance ($).

T. fremonti T. hudsonicus

0.911 0.9050.047 0.012

12.7⁄

31.3#,$

13.550.0⁄,$

15.8# 21.5

178 A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182

4. Discussion

This comprehensive phylogenetic assessment of the NorthAmerican red squirrels provides a more complex interpretationof speciation within these forest obligate species and characterizesprocesses that also influence other species of this increasingly frag-mented community. Congruence across species hints at fundamen-tal ecogeographic variability on a continental scale that haspredictable evolutionary consequences. Three main dynamicsemerge from our analyses.

The first dynamic reflects evolutionary complexity across NorthAmerica among forest dwelling species, with a dichotomy thatcontrasts minimal evolutionary structure in eastern North America(a single widespread lineage for red squirrels) with comparativelycomplex structure in the west (three species and multiple lin-eages). This east/west split reflects the Mississippi discontinuityas now reported across multiple other taxa (e.g., Soltis et al.,2006), with the Great Plains Ecoregion serving as a major barrierbetween east and west, and a gradient of increasing genetic com-plexity found within temperate forested biomes from east to west.Within the west, three geographic centers of diversity exist asreflected by T. douglasii in westernmost generally mesic forests,T. fremonti distributed through the Southwest, and T. hudsonicuswith an extensive distribution through remaining interconnectedforests.

The second dynamic reflects regional variability in spatialresponses of species to long-term climate trends. Both geneticdemographic analyses and niche predictions indicate relative sta-bility through the southwest region with over 60% of current nichefor T. fremonti consistently predicted as suitable through all recent(Late Quaternary) climate phases. By contrast, all lineages within T.hudsonicus except for the Central Rockies exhibit significant signalsof recent demographic expansion and virtually no persistent nichespace through timeframes. Dramatically different histories for T.hudsonicus and T. fremonti reflect alternate modes of diversifica-tion. In the southwest, T. fremonti is the product of ancient peri-patric isolation and subsequent divergence accompanied byperiodic fragmentation/reconnection at local scales. In contrast,glacial and interglacial phase distributions of T. hudsonicus arenon-overlapping, and such vast spatial displacement likely reflectsrepeated population extirpations as leading edge expansions andcontractions predominate through time, resulting in species-widerange shifts. Across the continent-wide distribution of T. hudsoni-cus, broad-scale fragmentation and range contractions with subse-quent differentiation in multiple isolated areas, generated highoverall genetic diversity. Tamiasciurus douglasii exhibits an inter-mediate history with reduced glacial–interglacial phase displace-ment and medium-term persistence of populations.

The third major dynamic is relatively rapid radiation of threespecies despite repeated and ongoing gene flow, reflecting againhigh regional complexity in western North America and climatephase expansion–contraction dynamics. Similar episodes of rapiddiversification are being uncovered in other forest-dwelling mam-mals with shared geographic distributions (e.g., Shafer et al., 2010).For instance, both the Sorex cinereus complex and Sorex palustriscomplex exhibit longitudinally distributed lineages that reflect aMississippi discontinuity, rapid speciation, high genetic complex-ity in the west, and ongoing gene flow among species (Hopeet al., 2012, 2014a). Similarities in diversification processes are alsofound among other sciurids. Chipmunks (multiple genera;Patterson and Norris, 2015) have a deeper history of speciationthan red squirrels and provide a model for examining how geneflow has impacted the formation of species (Good et al., 2008;Sullivan et al., 2014). In addition, both intra- and inter-specific dis-tributional histories among chipmunks reflect low diversity in

eastern North America (Tamias striatus and a single lineage of Neo-tamias minimus), as opposed to 22 species that have arisen in west-ern regions (Hall, 1981) with extensive hybridization at rangeboundaries. Marmots (Genus Marmota) also exhibit inter-specificgene flow (Kerhoulas et al., 2015) and the history of diversificationof multiple genera of ground squirrels has been accompanied byperiodic episodes of introgressive hybridization (Helgen et al.,2009). Recent speciation, coupled with replicated episodes of geneflow, both spatially and temporally, among North American redsquirrels should provide novel perspectives on the formation ofNearctic species (Abbott et al., 2013).

4.1. Systematic relationships

Given the extensive range of red squirrels, a comprehensiveassessment of systematic relationships and evolutionary historywithin Tamiasciurus has been hampered by limited geographicsampling and associated multi-locus genetic perspectives. Hall(1981) identified substantial morphological variation, recognizing4 subspecies within T. douglasii and 25 subspecies within T. hud-sonicus. Subsequently, Lindsay (1981) raised T. d. mearnsi to speciesstatus based on distinctive morphospace among the three species,but lacking range-wide sampling. The first geographically wide-spread genetic assessment based on a single locus (Arbogastet al., 2001) found no genetic divergence between T. douglasiiand T. mearnsi and instead recovered three independent lineagesas western (combined T. douglasii and T. mearnsi), eastern (all T.hudsonicus except for Southwestern populations) and Southwest-ern (Arizona, New Mexico, and southern Colorado populations ofT. hudsonicus). Additional mtDNA sequence data identified a para-phyletic relationship of T. hudsonicus with respect to T. douglasii(including T. mearnsi) and reported a distinct lineage located inthe Alexander Archipelago of Southeast Alaska (Barber, 2007).Most recently, multiple nuclear loci suggest refugial isolation onVancouver Island (Chavez et al., 2014), and indicate a complex his-tory of diversification including gene flow, as has been found else-where between T. hudsonicus and T. douglasii (Chavez et al., 2011).

Considering differentiation across multiple loci, climatic nichedivergence, and previous morphological evidence, the taxonomyof red squirrels should be updated. The isolated population occur-ring in Baja California (previously T. mearnsii) is minimally diver-gent from the remainder of the Western lineage over all lociexamined; however, a combined western clade representing theDouglas squirrel (T. douglasii and T. mearnsii) forms one of threespecies that are well supported over all analyses. The other twospecies are the southwestern red squirrel (T. fremonti; includingSouthern Rockies, Sacramento Mountains, and Southwestern SkyIsland lineages), and the broadly distributed American red squirrel(T. hudsonicus) that includes five mtDNA lineages (Continental,Northwestern, Central Rockies, North Pacific Coast, and VancouverIsland). These three species better reflect the evolutionary historyand ecology of boreal forest obligates and should improve manage-ment and conservation planning.

Although our methods allow for more rigorous representationof systematic relationships within Tamiasciurus, the revised taxon-omy does not fully account for either incomplete lineage sorting orrecent and historic admixture. In this regard, a closer appraisal ofthe nuclear evidence compared with matrilineal inheritance isinformative (Fahey et al., 2014; Hung et al., 2012). Three mtDNAlineages are highly supported as unique among red squirrels,including Vancouver Island, Central Rockies, and NPC lineages;however, all exhibit strong evidence of introgression across differ-ent temporal scales. Vancouver Island has fixed mito-nuclear dis-cord where the mitochondria are indicative of T. douglasii, butonly three haplotypes among nuclear loci examined are private

https://www.researchgate.net/publication/234162796_Hybridization_and_speciation?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/51504048_Genetic_and_phenotypic_variation_across_a_hybrid_zone_between_ecologically_divergent_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/259386029_Diversification_and_gene_flow_in_nascent_lineages_of_island_and_mainland_North_American_tree_squirrels_Tamiasciurus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/261675085_DNA-based_approaches_for_evaluating_historical_demography_in_terrestrial_vertebrates?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/253546310_Generic_Revision_in_the_Holarctic_Ground_Squirrel_Genus_Spermophilus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/253546310_Generic_Revision_in_the_Holarctic_Ground_Squirrel_Genus_Spermophilus?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/225085616_A_climate_for_speciation_Rapid_spatial_diversification_within_the_Sorex_cinereus_complex_of_shrews?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/225085616_A_climate_for_speciation_Rapid_spatial_diversification_within_the_Sorex_cinereus_complex_of_shrews?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/265462716_Multilocus_phylogeography_and_systematic_revision_of_North_American_water_shrews_genus_Sorex?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/230795881_Multilocus_coalescence_analyses_support_a_mtDNA-based_phylogeographic_history_for_a_widespread_Palearctic_passerine_bird_Sitta_europaea?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/280633004_Complex_history_of_isolation_and_gene_flow_in_hoary_Olympic_and_endangered_Vancouver_Island_marmots?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/277311512_Towards_a_uniform_nomenclature_for_ground_squirrels_The_status_of_the_Holarctic_chipmunks?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/6691230_Comparative_phylogeography_of_unglaciated_eastern_North_America?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/6691230_Comparative_phylogeography_of_unglaciated_eastern_North_America?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/261997310_Divergence_with_gene_flow_within_the_recent_chipmunk_radiation_Tamias?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/46303799_Shafer_ABA_Cullingham_CI_Cote_SD_Coltman_DW_Of_glaciers_and_refugia_a_decade_of_study_sheds_new_light_on_phylogeography_of_northwestern_North_America_Mol_Ecol_19_4589-4621?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/27397977_The_Mammals_of_North_America_Vol_1?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==https://www.researchgate.net/publication/null?el=1_x_8&enrichId=rgreq-59af66fbf18b5421b06ee762e8f08d5e-XXX&enrichSource=Y292ZXJQYWdlOzMwMTI3NzIyNztBUzozNTM3MTg3NzcyMDQ3MzdAMTQ2MTM0NDUzNjk1MA==

A.G. Hope et al. /Molecular Phylogenetics and Evolution 100 (2016) 170–182 179

(most being shared with T. hudsonicus). Given the single mtDNAlineage present on Vancouver Island, we concur with Chavezet al. (2014) that this population represents a refugial lineage thatpersisted through at least part of the last glacial, resulting from asmall founder event by h