Large‐Scale Migrations of Brown Bears in Eurasia and to North America During the Late Pleistocene

DOI: 10.1111/jbi.13126

ORIGINAL ARTICLE

Large-scale migrations of brown bears in Eurasia and to North America during the Late Pleistocene

1Department of Zoology, Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia 2School of Life and Environmental Sciences, University of Sydney, Sydney, NSW, Australia 3West Branch, Russian Research Institute of Game Management and Fur Farming, Saint Petersburg, Russia 4Russian Research Institute of Game Management and Fur Farming, Kirov, Russia 5N.K. Koltzov Institute of Developmental Biology, Russian Academy of Sciences, Moscow, Russia 6Institute of Plant and Animal Ecology, Ural Branch of Russian Academy of Sciences, Yekaterinburg, Russia 7Federal Scientific Center of the East Asia Terrestrial Biodiversity, Far Eastern Branch of the Russian Academy of Sciences, Vladivostok, Russia 8Natural Resources Institute, Rovaniemi, Finland 9Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of Life Sciences, As, Norway 10Norwegian Institute for Nature Research, Trondheim, Norway 11NIBIO, Norwegian Institute of Bioeconomy Research, Svanvik, Norway 12Faculty of Forestry, Technical University, Zvolen, Slovakia

Correspondence Urmas Saarma, Department of Zoology, Abstract Institute of Ecology and Earth Sciences, Aim: Climatic changes during the Late Pleistocene had major impacts on popula- University of Tartu, Tartu, Estonia. Email: [email protected] tions of plant and animal species. Brown bears and other large mammals are likely to have experienced analogous ecological pressures and phylogeographical pro- Funding information Seventh Framework Programme, Grant/ cesses. Here, we address several unresolved issues regarding the Late Pleistocene Award Number: contracts 247652 and demography of brown bears: (1) the putative locations of refugia; (2) the direction 2096/7. PR UE/2011/2; Estonian Ministry of Education and Research, Grant/Award of migrations across Eurasia and into North America; and (3) parallels with the Number: IUT20-32 demographic histories of other wild mammals and modern humans.

Editor: Gerald Schneeweiss Location: Eurasia and North America. Methods: We sequenced 110 complete mitochondrial genomes from Eurasian brown bears and combined these with published sequences from 138 brown bears and 33 polar bears. We used a Bayesian approach to obtain a joint estimate of the phylogeny and evolutionary divergence times. The inferred mutation rate was compared with estimates obtained using two additional methods. Results: Bayesian phylogenetic analysis identified seven clades of brown bears, with most individuals belonging to a very large Holarctic clade. Bears from the widespread clade 3a1, which has a distribution from Europe across Asia to Alaska, shared a common ancestor about 45,000 years ago.

Main conclusions: We suggest that the Altai-Sayan region and Beringia were important Late Pleistocene refuge areas for brown bears and propose large-scale migration scenarios for bears in Eurasia and to North America. We also argue that brown bears and modern humans experienced a demographic standstill in Beringia before colonizing North America.

KEYWORDS Altai, Beringia, climate change, mitochondrial genome, molecular clock, phylogeography, Ursus arctos

1 | INTRODUCTION diversity and demography. Important insights into geographical variation among populations of animals, such as those of the brown bear, Climatic fluctuations during the Pleistocene have shaped the distri- have been gained through analyses of autosomal loci (Swenson bution, genetic structure, and diversity of present-day species in Eur- et al., 2011; Tammeleht et al., 2010), Y-chromosome markers (Bidon asia and North America. During the Last Glacial Maximum (LGM; et al., 2014; Schregel et al., 2015), and other sources of data, such 26.5–19 kya), many high-latitude areas were covered by continental as diet (Vulla et al., 2009). Based on complete mitogenome ice sheets and temperate species were largely restricted to warmer sequences of brown bears in north-western Eurasia, Keis et al. and more hospitable areas known as glacial refugia (Hofreiter & Ste- (2013) identified population structuring, demographic history, and wart, 2009). Some southern parts of Europe, such as the Mediter- spatial population processes that had not previously been detected ranean peninsulas, have long been recognized as LGM refugia for using shorter sequences. temperate species. There is also growing recognition of more north- We expect that a wider analysis of mitogenomes from Eurasia to erly refugia in Europe, including the Carpathian Mountains (Def- North America will help to expose new details of ancient population fontaine et al., 2005; Kotlık et al., 2006; Saarma et al., 2007; Schmitt processes and their time-frame. Mitochondrial DNA has been a pre- & Varga, 2012). ferred genetic marker for studying bears for several reasons. Perhaps In phylogeographical studies of large mammals, the brown bear the most relevant to phylogeographical studies is that because (Ursus arctos L.) has been a model species. This is largely due to its female brown bears are philopatric, the phylogeographical signal is wide past and present distribution, as well as female philopatry (Davi- better preserved in mitochondrial DNA than in autosomal markers or son et al., 2011). In historical times, brown bears were common in the Y-chromosome. However, the phylogeographic history of the most parts of Eurasia and in the western part of North America down female lineage does not necessarily coincide with that of the whole to northern Mexico (Kurten, 1968). Present-day bear populations species (Davison et al., 2011). A notable example of this is the mito- occupy somewhat smaller areas, with many populations fragmented chondrial paraphyly of brown bears (Talbot & Shields, 1996), which mainly because of human persecution and loss of suitable habitat stands in contrast with the clear support for monophyly from male- (Davison et al., 2011; Swenson, Taberlet, & Bellemain, 2011). The lar- specific (Bidon et al., 2014) and biparental markers (Liu et al., 2014). gest brown bear population in northern Eurasia stretches from Scan- The present-day European population of brown bears consists of dinavia to the Russian Far East and is associated predominantly with two main mitochondrial lineages, the eastern (clade 3a) and western the occurrence of boreal and temperate forests. The range is (clade 1), of which the latter is divided into subclades 1a and 1b restricted by tundra in the north and by dry steppe in the south (Taberlet & Bouvet, 1994). These clades and subclades exhibit con- (Servheen, Herrero, & Peyton, 1999). However, the Gobi bear in siderable geographical differentiation. Subclade 1a originated from Mongolia is desert-dwelling (McCarthy, Waits, & Mijiddorj, 2009), an Iberian refuge and is now found in southwestern Europe and indicating that brown bears can survive in fairly hostile environments. southern Scandinavia; however, the recolonization of southern Scan- For many species, including the brown bear, long-term isolation dinavia was most likely not limited to migrations from a single refu- of populations in separated glacial refugia resulted in genetic diver- gium (Bray et al., 2013). In contrast, subclade 1b has virtually gence among populations, loss of genetic variability and complex remained in its refugial territory in the south and east of Europe patterns of migrations during the Late Pleistocene and Holocene (Italy and Balkans). In North America, four mitochondrial clades have (Davison et al., 2011; Hewitt, 2000; Taberlet, Fumagalli, Wust-Saucy, been identified in contemporary brown bears: clade 2a on the Admi- & Cosson, 1998). However, despite the wide Eurasian distributions ralty, Baranof, and Chichagof (ABC) islands of Alaska, clade 3a1 in of many species, including several model species for biogeography, western Alaska, clade 3b in eastern Alaska, and clade 4 in the south- we lack an equivalent understanding of the processes influencing ern distribution of the species in continental North America (Davison populations in Asia. et al., 2011; Hirata et al., 2013). Analysis of genetic markers has provided a highly informative The eastern clade present in Europe (3a) divides further into sub- means of reconstructing both ancient and contemporary patterns of clades 3a1 and 3a2 (Hirata et al., 2013). Subclade 3a1 extends from 396 | ANIJALG ET AL.

Fennoscandia to East Asia, and into North America (Keis et al., publicly available data from Eurasia to North America in a dated phy- 2013; Korsten et al., 2009; Murtskhvaladze, Gavashelishvili, & Tarkh- logenetic analysis. We hypothesized that: (1) the widely distributed nishvili, 2010; Saarma & Kojola, 2007; Waits, Talbot, Ward, & clade 3a1 would display Eurasia-wide structure that has not been Shields, 1998). Other clades in Asia appear geographically rather apparent using lower-resolution genetic markers; and (2) populations confined: clade 3a2 to central Hokkaido, 3b to eastern Hokkaido, were restricted to refugia in the areas of Altai, Beringia and the the Russian Far East and southern Siberia (Tomsk region), clade 4 to Urals, such that the ages of the clades containing individuals from southern Hokkaido, and clade 5 to Tibet (Guskov, Sheremeteva, these areas are consistent with the timing of glacial restriction. We Seredkin, & Kryukov, 2013; Hirata et al., 2013; Salomashkina, Kholo- also compared our data with existing information from modern dova, Tutenkov, Moskvitina, & Erokhin, 2014). humans and other mammal species to assess the generality of Late Although broad patterns of mitogenomic composition are known Pleistocene migration scenarios. for Asian brown bear populations, phylogeographical syntheses of these patterns remain tentative. The widespread sublineage 3a1 is believed to have emerged from multiple glacial refugia, but the loca- 2 | MATERIALS AND METHODS tions of putative areas in Asia remain unknown. The Siberian plains and mountains were largely unglaciated during the LGM, represent- Tissue samples from 110 brown bears, legally harvested between ing potential refugia (Davison et al., 2011). Indeed, several studies 1999 and 2010 for purposes other than this study, were used for have suggested that the Altai-Sayan region in south-central Siberia full mitochondrial DNA sequencing. These included three from was one of the most important Asian refugia during the Late Pleis- Romania, five from Estonia, five from Sweden, 10 from Finland and tocene, due to its stable climatic conditions and species richness 87 from the Russian Federation (Figure 1). For DNA extraction, we throughout this period of time (Pavelkova Ri cankova, Robovsky, & used the High PCR Template Preparation Kit (Roche Diagnostics, Riegert, 2014; Pavelkova Ri cankova, Robovsky, Riegert, & Zrzavy, Mannheim, Germany) following the manufacturer’s instructions. 2015). Moreover, modern humans are known to have inhabited the PCR and sequencing of the complete mitochondrial genomes Altai region for at least 45 kyr (Goebel, 1999) and the area played a was done as described by Keis et al. (2013), with a modified set of significant role in the peopling of Siberia and North America (Dulik primers; we replaced six primers with newly designed ones to et al., 2012). Reconstructing the migrations of modern humans dur- improve mitogenome assembly (Table S1 in Appendix S1). All ing the Late Pleistocene and early Holocene (c. 50–10 kya) poses a sequences have been deposited in GenBank (accession numbers major challenge, because of the small population size and low den- KY419593–KY419702). For phylogenetic analysis, we added sity of archaeological sites during this period. However, other large sequences from 138 brown bears and 33 polar bears (U. maritimus) mammal species that shared similar ecological demands as modern from GenBank (Table S4 in Appendix S1). In total, our data set con- humans, such as the brown bear, wapiti (Cervus canadensis; Meiri tained 281 mitogenome sequences from brown bears and polar et al., 2014) and grey wolf (Canis lupus; Koblmuller€ et al., 2016), rep- bears. resent a largely unexploited source of information that can also To estimate the matrilineal phylogeny and evolutionary time- improve our understanding of the drivers of ancient population pro- scale of brown bears, we analysed complete mitogenomes using a cesses in modern humans. Bayesian approach in BEAST 1.8.4 (Drummond, Suchard, Xie, & Ram- Several other geographical regions are of potential relevance for baut, 2012). The best-fitting partitioning scheme was chosen based understanding genetic diversity in Asia and links with the adjacent on the Bayesian information criterion using PartitionFinder (Lanfear, regions of Europe and North America. First, there is evidence to sug- Calcott, Ho, & Guindon, 2012), which partitioned the alignment gest that the Ural Mountains acted as a glacial refugium for animals into six subsets: D-loop, rRNA genes, tRNA genes, and the three and modern humans at the border of Europe and Asia (Svendsen codon positions of all 13 protein-coding genes. We compared con- et al., 2010). Another refugium is believed to have been located in stant-size and skyride coalescent models for the tree prior using

Beringia, the area stretching from the Verkhoyansk Mountains and Bayes factors in BEAST, calculated using the stepping-stone estimator Lena River Basin in eastern Eurasia to Alaska and the Northwest of the marginal likelihood (Xie, Lewis, Fan, Kuo, & Chen, 2011). Territories of Canada in North America. Compared with areas of sim- The constant-size model was decisively favoured (BF > 100) and ilar latitude, the climate in Beringia stayed relatively mild and ice-free we used this to provide the prior for the tree topology and relative during the LGM due to the warm North Pacific circulation (Lambeck, node times. Yokoyama, & Purcell, 2002). The Bering Land Bridge, which emerged We analysed the partitioned alignment using a strict molecular during times of low sea level, allowed migrations of mammals clock, owing to the intraspecific nature of the data. Preliminary anal- between Eurasia and North America several times during the Pleis- yses using a more parameter-rich uncorrelated lognormal relaxed tocene. clock led to poor mixing and convergence in our analyses. A rate The aim of this study was to investigate the Late Pleistocene multiplier was specified for each data subset. The molecular clock phylogeographical history of brown bears in Eurasia and their colo- was calibrated by the inclusion of a sequence from an ancient polar nization of North America. We sequenced the complete mitochon- bear (age 120,000 years; GU573488). Estimates of posterior distri- drial genomes of 110 brown bears across Eurasia and included butions were obtained via Markov chain Monte Carlo sampling, with ANIJALG ET AL. | 397

FIGURE 1 Map showing locations of Eurasian and North American brown bears analysed in this study: 110 newly sequenced mitogenomes (circles) and 138 from databases (triangles). Colours correspond to clades in Figure 2 [Colour figure can be viewed at wileyonlinelibrary.com] samples drawn every 5,000 steps over a total of 50,000,000 steps. 3 | RESULTS Each analysis was run in duplicate and the samples were combined after removing the first 10% as burn-in. We checked that the effec- Our Bayesian phylogenetic analysis of 281 mitogenome sequences tive sample sizes of all parameters were >200. from brown bears to polar bears distinguished eight clades (Fig- We performed a date-randomization test to investigate the ure 2). According to the prevailing terminology (Davison et al., 2011; performance of the sequence ages as calibrations for the molecu- Hirata et al., 2013), these clades are: 1 and 2a (and 2b for polar lar clock (Ramsden, Holmes, & Charleston, 2009) and generated bears) representing the western lineage; and 3a1, 3a2, 3b, 4 and 5 10 replicate data sets in which the sequence ages were randomly representing the eastern lineage. All newly sequenced samples fell reassigned. The mean rate estimate from the original data set was into the widely distributed clade 3a1, except for one sample from not included in the 95% credibility intervals (CI) of the rate esti- Sweden which belonged to western clade 1. mates from any of the 10 date-randomized replicates. Further- In our Bayesian phylogenetic analysis, the mitogenomic mutation more, the rate estimates from the date-randomized replicates had rate was estimated at 2.48 9 10À8 mutations per site per year, with a long lower tails that spanned multiple orders of magnitude. In 95% CI of 1.93 9 10À8 to 3.04 9 10À8 mutations per site per year. combination, these features suggest that the original data set has This estimate was slightly lower than the estimate based on regression sufficient temporal structure and spread for calibrating the esti- of root-to-tip distances against sample age (3.35 9 10À8 mutations mate of the rate (Duchene,^ Duchene,^ Holmes, & Ho, 2015; Ho per site per year), but similar to the rate estimated using least-squares et al., 2011). dating (2.84 9 10À8 mutations per site per year). As an independent check on the reliability of the rate estimate, The most recent common ancestor (MRCA) of clade 3a1 lived we analysed the data using two other tip-dating methods: root-to- c. 45 kya (95% CI: 31–58 kya) (Figure 2; see also Table S2 and tip regression and least-squares dating. For both of these methods, Table S3 for 95% CIs, in Appendix S1). This split separated brown we inferred a phylogenetic tree using maximum likelihood with 100 bears currently inhabiting Romania and Bulgaria from the rest of the random starts in RAxML 8.2.4 (Stamatakis, 2014). The inferred phy- 3a1 clade. The latter diverged 37 kya (95% CI: 27–48 kya) into two logram was then used for regression of root-to-tip distances against large subclades: (1) a Kamchatkan–Alaskan subclade comprising bears sampling time in TempEst (Rambaut, Lam, Carvalho, & Pybus, 2016) from Alaska and Kamchatka (34 kya, 95% CI: 24–44 kya); and (2) a and for least-squares dating in LSD (To, Jung, Lycett, & Gascuel, Eurasian subclade that includes bears inhabiting the vast area from 2016). eastern Europe to the Russian Far East (32 kya, 95% CI 23–42 kya). 398 | ANIJALG ET AL.

FIGURE 2 Bayesian phylogenetic tree of brown bears inferred from mitochondrial genomes. The numbers on nodes and after clades represent posterior mean estimates of the ages of different bear lineages, with numbers in parentheses representing the 95% credibility intervals. Red circles at nodes represent posterior probabilities >0.95. The blue vertical columns represent long cold periods according to Ahn and Brook (2014). The tree is drawn to a time-scale, with node ages given in thousands of years before present. Colours indicate different mitochondrial clades. Colours of clades correspond to those in Figure 1 [Colour figure can be viewed at wileyonlinelibrary.com]

Following the formation of the Kamchatkan-Alaskan subclade, the phylogeographical processes affecting mammals in the vast terri- the Alaskan bears separated from the Kamchatkan bears (27 kya, tory of northern Eurasia. An important step towards improving this 95% CI 19–36 kya). The latter formed two clusters: “to E-Kam- knowledge is to gain a reliable estimate of the evolutionary rate and chatka” (14 kya, 95% CI 7–21 kya), consisting mostly of bears time-scale of each species (Ho, Saarma, Barnett, Haile, & Shapiro, inhabiting the north-eastern part of the peninsula; and “to W-Kam- 2008). Our analyses confirm previous suggestions that a single chatka” (25 kya, 95% CI 17–33 kya), largely represented by bears ancient DNA sequence can be sufficient for calibrating the molecular from the south-western part of the peninsula. The Eurasian subclade clock, provided that it is sufficiently old in comparison with the evo- divided into lineages that reached Primorye, Sakhalin, and Kam- lutionary time-scale of the samples in the data set (Molak, Lorenzen, chatka (27 kya, 95% CI 18–37 kya) and one that spread widely in Shapiro, & Ho, 2013). This is further supported by the consistency in Eurasia (29 kya, 95% CI 21–38 kya). The latter divided further into our estimates of the mutation rate using three different methods. two: bears that eventually reached Europe and bears that migrated To date, only shallow regional genetic structure has been to the Urals and to Magadan region, with only minimal overlap in detected among several mammal species occurring across northern the Perm and Sverdlovsk regions. Eurasia (e.g. Hindrikson et al., 2017; Korsten et al., 2009). However, that emerging picture is by no means definitive, as these compar- isons were made using relatively short mitochondrial sequences 4 | DISCUSSION (Davison et al., 2011). Based on our analyses of complete mitogenome data, we propose several scenarios of brown bear migration Although the phylogeography of European mammals has been stud- across Eurasia and to North America during the last 50 kyr ied fairly extensively, we have only a fragmentary understanding of (Figure 3). ANIJALG ET AL. | 399

FIGURE 3 Possible distribution and migration scenarios of brown bears from clade 3a1 in the Late Pleistocene. Colours correspond to those in Figures 1 and 2. Excerpts of the corresponding parts of the phylogenetic tree (according to Figure 2) are shown on the right of each panel. Subfossils are marked by large brown circles [Colour figure can be viewed at wileyonlinelibrary.com]

been wide. Two radiocarbon-dated 3a bear samples are known from 4.1 | Spatial distribution of bear clade 3a in the this period, both c. 47 ky old: one from Ramesch Cave, Austria Late Pleistocene (Hofreiter et al., 2004) and the other from Tain Cave in the Urals, The full distribution of clade 3a bears around 50 kya remains Russia (Bray, 2010). The distribution of clade 3a probably reached unknown, but existing data indicate that the distribution could have even farther east. One of the arguments to support this is the 400 | ANIJALG ET AL. divergence of 3a1 and 3a2 that started c. 53 kya (95% CI: 38– and also for other species, such as the common vole (Microtus arva- 71 kya). As 3a2 bears are found only in Hokkaido (Hirata et al., lis) (Stojak, McDevitt, Herman, Searle, & Wojcik, 2015) and weasel 2013), the likely distribution of 3a bears before the split extended to (Mustela nivalis) (McDevitt et al., 2012). Two other Romanian sam- eastern Asia. The vast distribution is also supported by the biotope ples belonged to the much larger second lineage of 3a1 (Figure 2), map in Hoogakker et al. (2016) which shows that around 54 kya which moved across Eurasia and to North America. This suggests boreal and temperate forests covered a continuous area from west- that bears made their way to the Carpathians (Romania) not only ern Europe to the Russian Far East (Figure 4c). The wide distribution from the north (central Europe), but also from the east (from the of the 3a lineage might have been ruptured by significant global cli- Altai-Sayan region, discussed below). mate cooling about 48–54 kya (Kindler et al., 2014), with these The clade that spread throughout Eurasia and part of North bears potentially restricted to different refugia (Figure 3a). America shared a most recent common ancestor about 37 kya (95% The most recent common ancestor of clade 3a1 lived 45 kya CI: 27–48 kya; Figure 3b). Several lines of evidence suggest that this (95% CI: 31–58 kya). This clade split into two mitochondrial lineages, ancestral population was restricted to a refuge area in the mountain- the first of which is represented by bears from Romania to Bulgaria. ous Altai-Sayan region of Central Asia. This region is known as a bio- These might be descendants of ancient bears that lived in central diversity hotspot for temperate flora and fauna, having maintained Europe c. 50 kya (e.g. the individual from Ramesch Cave) that found high ecological stability since the Late Pleistocene (Allen et al., 2010; refuge south of the Carpathian Mountains. The Carpathian region Huntley et al., 2013). Detailed analysis of the Late Pleistocene mam- probably acted as a glacial refugium for brown bears, as previously mal assemblages in the Altai region has shown that no significant suggested by Sommer and Benecke (2005) and Saarma et al. (2007), changes in composition or ecological structure occurred between the

FIGURE 4 Maps showing the extent of different biotopes (a) 6 kya, (b) 21 kya and (c) 54 kya (modified from Hoogakker et al., 2016). Note that the precise distribution of biotypes – for example the extent of forest in central Eurasia in periods (b) and (c) – comes with a degree of uncertainty (see Hoogakker et al., 2016 for discussion) Potential refuge areas are shown with dashed lines [Colour figure can be viewed at wileyonlinelibrary.com] ANIJALG ET AL. | 401

Late Pleistocene and the Holocene, providing a modern analogue for describing the LGM (Figure 4b) shows that central Beringia was cov- the Pleistocene communities (Agadjanian & Serdyuk, 2005; ered by boreal forest, providing suitable habitat for brown bears. Pavelkova Ri cankov a et al., 2014, 2015). There has been growing genetic support for the standstill hypothesis The Altai-Sayan region is at the geographical centre of the brown (Watson, 2017), including evidence from ancient mitochondrial gen- bear 3a1 distribution, making it the most parsimonious location for omes from modern humans (Llamas et al., 2016). A study of wapiti the ancestral refuge area. Furthermore, historical bear samples exhi- also found that this species was present in western Beringia bit higher genetic diversity in this region than elsewhere in northern c. 50 kya, but did not advance to North America until 15 kya (Meiri Eurasia (Hirata, Abramov, Baryshnikov, & Masuda, 2014). The Altai- et al., 2014). Sayan region is likely to have served as an important glacial refugium The favourable conditions for many plant and animal species in for modern humans and as an important route for the peopling of western and central Beringia during the LGM are believed to have northern Asia (Derenko et al., 2014). been caused by the North Pacific circulation, which brought warm and moist air to the area (Brubaker, Anderson, Edwards, & Lozhkin, 2005; Hoffecker et al., 2014; Willerslev et al., 2014). However, such 4.2 | Colonizing North America good conditions were not available in eastern Beringia before One of the major subclades of 3a1 bears headed towards Beringia, 15 kya. Here the environment was apparently a cold, semi-arid, and eventually colonizing Kamchatka and Alaska. Bears from the other treeless steppe-tundra, suitable only for species adapted to extreme major subclade of 3a1 dispersed widely throughout Eurasia. Interest- cold, such as woolly mammoths (Mammuthus primigenius) (Guthrie, ingly, bears in the Magadan area, a region close to Kamchatka, were 2001). neither part of the migration wave that colonized the Kamchatka Synchrony in the timing of colonization and expansion implies Peninsula, nor part of the wave that entered North America. Rather, that the ecology of large mammals, such as the wapiti and brown these bears apparently did not move farther north-east. However, bear, includes key limiting factors that can enhance our understand- we found a bear from the east Asian migration in Kamchatka (Fig- ing of the ancient movements of modern humans (Meiri et al., ure 3c), suggesting that some bears from this population eventually 2014). However, despite similarities between the migrations of reached Kamchatka, perhaps somewhat later than bears that were brown bears and modern humans towards North America c. 15 kya, part of the earlier migration that colonized North America. there are also some important differences. Brown bears are known Brown bears belonging to the Kamchatkan-Alaskan mitochondrial to have colonized North America earlier; clades 2c, 3c, and 4 were subclade apparently entered western Beringia before the beginning already in North America before the last glaciation (Barnes, Matheus, of the LGM, because north-eastern territories outside Beringia were Shapiro, Jensen, & Cooper, 2002; Davison et al., 2011). The genetic mainly polar desert during the LGM and unsuitable for brown bears diversity of Beringian brown bears in the pre-glacial period (Ray & Adams, 2001). Indeed, Pitulko et al. (2004) described c. 34- (c. 40 kya) was higher than that of modern North American popula- kyr-old brown bear remains at the Yana site (Yana RHS), whereas tions, but by 15 kya the genetic diversity was similar to that of con- Bray (2010) was able to sequence a 135-bp mitochondrial fragment temporary bears in the New World (Leonard, Wayne, & Cooper, of a c. 30-kyr-old brown bear sample from Indigirka area that 2000). belonged to clade 3a. Both sites are in western Beringia, providing evidence that brown bears were present there before the LGM. 4.3 | Colonizing the Kamchatka Peninsula Moreover, our data suggest that the Kamchatkan-Alaskan subclade also reached western Beringia around that time (see discussion The extent and timing of the last glaciation in Kamchatka were lar- about Kamchatka below). There is no biotope map relating to this gely influenced by the Laurentide Ice Sheet (Barr & Solomina, 2014). time period, but the 54 kya map shows an almost continuous distri- The ice is believed to have advanced in two major stages in Late bution of boreal forest from Eurasia to central Beringia (Figure 4c). Pleistocene Kamchatka: the first advance was extensive and Around 37 kya this distribution was probably interrupted by abrupt occurred between 60 and 31 kya; the more recent advance was less climate change, thus separating the Kamchatkan-Alaskan subclade. extensive, characterized primarily by mountain glaciation, and coin- The MRCA of the Alaskan brown bears lived 22 kya (95% CI: cided with the LGM. The MRCA of brown bears that colonized 13–32 kya; Figure 2). The most likely location of these bears was North America and Kamchatka was dated to 34 kya (95% CI: 24– west of the current Bering Strait, where they remained for about 6– 44 kya), coinciding with the end of the major glaciation on Kam- 7 thousand years (Beringian Standstill), before crossing the Bering chatka. At this point, bears would have been able to occupy the ter- Land Bridge and reaching Alaska c. 15 kya. The Beringian Standstill ritory in the vicinity of the peninsula, or even enter the peninsula to hypothesis was originally proposed for migrations of modern some extent. The biotope map (Figure 4b) shows that boreal forests humans, whereby the ancestors of Native Americans remained in were present to a limited extent in northern Kamchatka during the western Beringia for thousands of years before migrating to North LGM, making it a suitable habitat for brown bears that later colo- America (Tamm et al., 2007). During this time the climatic conditions nized the peninsula. were unsuitable for modern humans at higher latitudes outside the Kurten (1968) mentioned that the American broad-skulled bears land bridge (Hoffecker, Elias, & O’Rourke, 2014). The biotope map morphologically resemble the Kamchatkan bears, whereas the 402 | ANIJALG ET AL.

American narrow-skulled bears are more similar to the bears from help. We also thank Eline Lorenzen for providing sequence data. This north-eastern Siberia. The morphological resemblance might be the work was supported by the institutional research funding (grant result of diet, because the largest bears occur in regions where they IUT20-32) from the Estonian Ministry of Education and Research; have access to large amounts of meat, especially salmon (Hilderbrand BIOGEAST – Biodiversity of East-European and Siberian large mam- et al., 1999). Another mammal species, the northern red-backed vole mals on the level of genetic variation in populations, 7th Framework (Myodes rutilus), exhibits a similar Kamchatkan-Alaskan clade (Kohli, Programme (contracts 247652 and 2096/7. PR UE/2011/2); Aus- Fedorov, Waltari, & Cook, 2015). The close phylogenetic association tralian Research Council; Swedish Environmental Protection Agency; between Alaskan and Kamchatkan populations in these two species Norwegian Environment Agency; Research Council of Norway; and points to a migration pattern that might also be shared by other the Finnish Ministry of Agriculture and Forestry. mammals.

ORCID 4.4 | Large-scale migrations across Siberia and Urmas Saarma http://orcid.org/0000-0002-3222-571X expansion to Europe

The MRCA of the Eurasian subclades was dated to 32 kya (95% CI: 23–42 kya). At this point, most of the Eurasian bears split from the REFERENCES branch that today can be found in the Primorye region, Sakhalin Agadjanian, A. K., & Serdyuk, N. V. (2005). The history of mammalian Island and Kamchatka, suggesting that at least some bears managed communities and paleogeography of the Altai Mountains in the Pale- to reach Kamchatka after the initial colonization of the peninsula olithic. Paleontological Journal, 39, 645–821. (Figure 3c). Bears from this subclade were probably restricted to a Ahn, J., & Brook, E. J. (2014). Siple Dome ice reveals two modes of millennial CO2 change during the last ice age. Nature Communications, 5, refuge area in the Far East during the LGM, a plausible region being 3723. https://doi.org/10.1038/ncomms4723 the Manchurian Plain. According to the climate model by Hoogakker Allen, J. R. M., Hickler, T., Singarayer, J. S., Sykesb, M. T., Valdesc, P. J., & et al. (2016), around 21 kya the Manchurian Plain was suitable for Huntley, B. (2010). Last glacial vegetation of northern Eurasia. Qua- boreal and temperate forests, making it potentially habitable for ternary Science Reviews, 29, 2604–2618. https://doi.org/10.1016/j. brown bears (Figure 4b). quascirev.2010.05.031 Barnes, I., Matheus, P., Shapiro, B., Jensen, D., & Cooper, A. (2002). The other Eurasian branch split into two around 29 kya (95% CI: Dynamics of Pleistocene population extinctions in Beringian brown 21–38 kya; Figure 3d). One of the lineages moved west towards the bears. Science, 295, 2267–2270. https://doi.org/10.1126/science. Ural Mountains and the other eastward. The Ural Mountains have 1067814 been suggested as a LGM refugium for several animal species, such Barr, I. D., & Solomina, O. (2014). Pleistocene and Holocene glacier fluctuations upon the Kamchatka Peninsula. Global and Planetary Change, as the field vole (Microtus agrestis) (Jaarola & Searle, 2002). The 113, 110–120. https://doi.org/10.1016/j.gloplacha.2013.08.005 MRCA of European bears lived around 25 kya (95% CI: 17–32 kya) Bidon, T., Janke, A., Fain, S. R., Eiken, H. G., Hagen, S. B., Saarma, U., ... Hai- and, as conditions improved, this lineage colonized eastern and ler, F. (2014). Brown and polar bear Y chromosomes reveal extensive northern Europe, reaching Romania (Figure 3d). The clade that male-biased gene flow within brother lineages. Molecular Biology and Evo- lution, 31, 1353–1363. https://doi.org/10.1093/molbev/msu109 remained east of the Ural Mountains started to diverge about Bray, S. C. E. (2010) Mitochondrial DNA analysis of the evolution and – 25 kya (95% CI: 17 34 kya; Figure 3d). One lineage migrated genetic diversity of ancient and extinct bears (PhD Thesis). University towards the Ural Mountains and the other towards Magadan region. of Adelaide. Although the western migration reached the Ural Mountains, this lin- Bray, S. C. E., Austin, J. J., Metcalf, J. L., Østbye, K., Østbye, E., Lauritzen, S.-E, ... Cooper, A. (2013) Ancient DNA identifies post-glacial eage has not been found farther west than the Perm region on the recolonisation, not recent bottlenecks, as the primary driver of con- western slopes of the Urals. As with the 3a1 clade in brown bears, temporary mtDNA phylogeography and diversity in Scandinavian the common juniper (Juniperus communis) also has a dominant and brown bears. Diversity and Distributions, 19, 245–256. https://doi. widespread clade in Eurasia (Hantemirova et al., 2017), although it org/10.1111/ddi.2013.19.issue-3 Brubaker, L. B., Anderson, P. M., Edwards, M. E., & Lozhkin, A. V. (2005). does not reach as far east. Beringia as a glacial refugium for boreal trees and shrubs: New per- In conclusion, our mitogenome analysis of brown bears in Eurasia spectives from mapped pollen data. Journal of Biogeography, 32, 833– and North America demonstrates that this species shared several 848. https://doi.org/10.1111/jbi.2005.32.issue-5 population processes in common with wapiti and modern humans Davison, J., Ho, S. Y. W., Bray, S. C., Korsten, M., Tammeleht, E., Hindrik- ... during the last 50 thousand years, including the Beringian Standstill. son, M., Saarma, U. (2011) Late-Quaternary biogeographic scenarios for the brown bear (Ursus arctos), a wild mammal model species. This suggests that wild mammal species and modern humans Quaternary Science Reviews, 30, 418–430. https://doi.org/10.1016/j. responded to the same key limiting factors in the Late Pleistocene. quascirev.2010.11.023 Deffontaine, V., Libois, R., Kotlık, P., Sommer, R., Nieberding, C., Par- adis, E., ... Michaux, J.R. (2005). Beyond the Mediterranean penin- ACKNOWLEDGEMENTS sulas: Evidence of central European glacial refugia for a temperate forest mammal species, the bank vole (Clethrionomys glareolus). The authors thank three anonymous reviewers for their insightful Molecular Ecology, 14, 1727–1739. https://doi.org/10.1111/mec. comments and suggestions, and Dr. Frank Zachos for his generous 2005.14.issue-6 ANIJALG ET AL. | 403

Derenko, M., Malyarchuk, B., Denisova, G., Perkova, M., Litvinov, A., Hoffecker, J. F., Elias, S. A., & O’Rourke, D. H. (2014). Out of Beringia? Grzybowski, T., & Zakharov, I. (2014). Western Eurasian ancestry in Science, 343, 979–980. https://doi.org/10.1126/science.1250768 modern Siberians based on mitogenomic data. BMC Evolutionary Biol- Hofreiter, M., Serre, D., Rohland, N., Rabeder, G., Nagel, D., Conard, N., ogy, 14, 217. https://doi.org/10.1186/s12862-014-0217-9 ... P€a€abo, S. (2004) Lack of phylogeography in European mammals Drummond, A. J., Suchard, M. A., Xie, D., & Rambaut, A. (2012). Bayesian before the last glaciation. Proceedings of the National Academy of phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Sciences of the United States of America, 101, 12963–12968. https://d Evolution, 29, 1969–1973. https://doi.org/10.1093/molbev/mss075 oi.org/10.1073/pnas.0403618101 Duchene,^ S., Duchene,^ D., Holmes, E. C., & Ho, S. Y. W. (2015). The per- Hofreiter, M., & Stewart, J. (2009). Ecological change, range fluctuations formance of the date-randomization test in phylogenetic analyses of and population dynamics during the Pleistocene. Current Biology, 19, time-structured virus data. Molecular Biology and Evolution, 32, 1895– R584–R594. https://doi.org/10.1016/j.cub.2009.06.030 1906. https://doi.org/10.1093/molbev/msv056 Hoogakker, B. A. A., Smith, R. S., Singarayer, J. S., Marchant, R., Prentice, Dulik, M. C., Zhadanov, S. I., Osipova, L. P., Askapuli, A., Gau, L., Gokcu- I. C., Allen, J. R. M., ... Tzedakis, C. (2016) Terrestrial biosphere men, O., ... Schurr, T. G. (2012) Mitochondrial DNA and Y chromo- changes over the last 120 kyr. Climate of the Past, 12,51–73. some variation provides evidence for a recent common ancestry https://doi.org/10.5194/cp-12-51-2016 between native Americans and indigenous Altaians. American Journal Huntley, B., Allen, J. R. M., Collingham, Y. C., Hickler, T., Lister, A. M., of Human Genetics, 90, 229–246. https://doi.org/10.1016/j.ajhg. Singarayer, J., ... Valdes, P.J. (2013) Millennial climatic fluctuations 2011.12.014 are key to the structure of Last Glacial Ecosystems. PLoS ONE, 8, Goebel, T. (1999). Pleistocene human colonization of Siberia and peopling e61963. https://doi.org/10.1371/journal.pone.0061963 of the Americas: An ecological approach. Evolutionary Anthropology, 8, Jaarola, M., & Searle, J. B. (2002). Phylogeography of field voles (Microtus 208–227. https://doi.org/10.1002/(ISSN)1520-6505 agrestis) in Eurasia inferred from mitochondrial DNA sequences. Guskov, V. Y., Sheremeteva, I. N., Seredkin, I. V., & Kryukov, A. P. (2013). Molecular Ecology, 11, 2613–2621. https://doi.org/10.1046/j.1365- Mitochondrial cytochrome b gene variation in brown bear (Ursus arc- 294X.2002.01639.x tos Linnaeus, 1758) from southern part of Russian Far East. Russian Keis, M., Remm, J., Ho, S. Y. W., Davison, J., Tammeleht, E., Tumanov, I. Journal of Genetics, 49, 1213–1218. https://doi.org/10.1134/ L., ... Saarma, U. (2013) Complete mitochondrial genomes and a S1022795413110070 novel spatial genetic method reveal cryptic phylogeographical struc- Guthrie, R. D. (2001). Origin and causes of the mammoth steppe: A story ture and migration patterns among brown bears in north-western of cloud cover, woolly mammal tooth pits, buckles, and inside-out Eurasia. Journal of Biogeography, 40, 915–927. https://doi.org/10. Beringia. Quaternary Science Reviews, 20, 549–574. https://doi.org/ 1111/jbi.2013.40.issue-5 10.1016/S0277-3791(00)00099-8 Kindler, P., Guillevic, M., Baumgartner, M., Schwander, J., Landais, A., & Hantemirova, E. V., Heinze, B., Knyazeva, S. G., Musaev, A. M., Lascoux, Leuenberger, M. (2014). Temperature reconstruction from 10 to 120 M., & Semerikov, V. L. (2017). A new Eurasian phylogeographical kyr b2k from the NGRIP ice core. Climate of the Past, 10, 887–902. paradigm? Limited contribution of southern populations to the recol- https://doi.org/10.5194/cp-10-887-2014 onization of high latitude populations in Juniperus communis L. Koblmuller,€ S., Vila, C., Lorente-Galdos, B., Dabad, M., Ramirez, O., Mar- (Cupressaceae). Journal of Biogeography, 44, 271–282. https://doi. ques-Bonet, T., ... Leonard, J. (2016) Whole mitochondrial genomes org/10.1111/jbi.2017.44.issue-2 illuminate ancient intercontinental dispersals of grey wolves (Canis Hewitt, G. (2000). The genetic legacy of the Quaternary ice ages. Nature, lupus). Journal of Biogeography, 43, 1728–1738. https://doi.org/10. 405, 907–913. https://doi.org/10.1038/35016000 1111/jbi.12765 Hilderbrand, G. V., Schwartz, C. C., Robbins, C. T., Jacoby, M. E., Hanley, Kohli, B. A., Fedorov, V. B., Waltari, E., & Cook, J. A. (2015). Phylogeog- T. A., Arthur, S. M., & Servheen, C. (1999). The importance of meat, raphy of a Holarctic rodent (Myodes rutilus): Testing high-latitude bio- particularly salmon, to body size, population productivity, and conser- geographical hypotheses and the dynamics of range shifts. Journal of vation of North American brown bears. Canadian Journal of Zoology, Biogeography, 42, 377–389. https://doi.org/10.1111/jbi.12433 77, 132–138. https://doi.org/10.1139/z98-195 Korsten, M., Ho, S. Y. W., Davison, J., Pahn,€ B., Vulla, E., Roht, M., ... Hindrikson, M., Remm, J., Pilot, M., Godinho, R., Stronen, A.V., & Bal- Saarma, U. (2009) Sudden expansion of a single brown bear maternal trunaite, L., ... Saarma, U. (2017) Wolf population genetics in Europe: lineage across northern continental Eurasia after the last ice age: A A systematic review, meta-analysis and suggestions for conservation general demographic model for mammals? Molecular Ecology, 18, and management. Biological Reviews, 92, 1601–1629. https://doi.org/ 1963–1979. https://doi.org/10.1111/j.1365-294X.2009.04163.x 10.1111/brv.12298 Kotlık, P., Deffontaine, V., Mascheretti, S., Zima, J., Michaux, J. R., & Hirata, D., Abramov, A. V., Baryshnikov, G. F., & Masuda, R. (2014). Mito- Searle, J. B. (2006). A northern glacial refugium for bank voles chondrial DNA haplogrouping of the brown bear, Ursus arctos (Car- (Clethrionomys glareolus). Proceedings of the National Academy of nivora: Ursidae) in Asia, based on a newly developed APLP analysis. Sciences of the United States of America, 103, 14860–14864. https://d Biological Journal of the Linnean Society, 111, 627–635. https://doi. oi.org/10.1073/pnas.0603237103 org/10.1111/bij.2014.111.issue-3 Kurten, B. (1968). Pleistocene mammals of Europe. London: Weidenfeld & Hirata, D., Mano, T., Abramov, A. V., Baryshnikov, G. F., Kosintsev, P. A., Nicolson. Vorobiev, A. A., ... Masuda, R. (2013) Molecular phylogeography of Lambeck, K., Yokoyama, Y., & Purcell, T. (2002). Into and out of the Last the brown bear (Ursus arctos) in Northeastern Asia based on analyses Glacial Maximum: Sea-level change during Oxygen Isotope Stages 3 of complete mitochondrial DNA sequences. Molecular Biology and and 2. Quaternary Science Reviews, 21, 343–360. https://doi.org/10. Evolution, 30, 1644–1652. https://doi.org/10.1093/molbev/mst077 1016/S0277-3791(01)00071-3 Ho, S. Y. W., Lanfear, R., Bromham, L., Phillips, M. J., Soubrier, J., Lanfear, R., Calcott, B., Ho, S. Y. W., & Guindon, S. (2012). PartitionFin- Rodrigo, A. G., & Cooper, A. (2011). Time-dependent rates of molecu- der: Combined selection of partitioning schemes and substitution lar evolution. Molecular Ecology, 20, 3087–3101. https://doi.org/10. models for phylogenetic analyses. Molecular Biology and Evolution, 29, 1111/j.1365-294X.2011.05178.x 1695–1701. https://doi.org/10.1093/molbev/mss020 Ho, S. Y. W., Saarma, U., Barnett, R., Haile, J., & Shapiro, B. (2008). The Leonard, J. A., Wayne, R. K., & Cooper, A. (2000). Population genetics of effect of inappropriate calibration: Three case studies in molecular Ice Age brown bears. Proceedings of the National Academy of Sciences ecology. PLoS ONE, 3, e1615. https://doi.org/10.1371/journal.pone. of the United States of America, 97, 1651–1654. https://doi.org/10. 0001615 1073/pnas.040453097 404 | ANIJALG ET AL.

Liu, S., Lorenzen, E. D., Fumagalli, M., Li, B., Harris, K., Xiong, Z., ... Biology Bulletin, 41,38–46. https://doi.org/10.1134/S10623590140 Wang, J. (2014). Population genomics reveal recent speciation and 10087 rapid evolutionary adaptation in polar bears. Cell, 157, 785–794. Schmitt, T., & Varga, Z. (2012). Extra-Mediterranean refugia: The rule and https://doi.org/10.1016/j.cell.2014.03.054 not the exception? Frontiers in Zoology, 9, 22. https://doi.org/10. Llamas, B., Fehren-Schmitz, L., Valverde, G., Soubrier, J., Mallick, S., Roh- 1186/1742-9994-9-22 land, N., ... Haak, W. (2016). Ancient mitochondrial DNA provides Schregel, J., Eiken, H. G., Grøndahl, F. A., Hailer, F., Aspi, J., Kojola, I, ... high-resolution time scale of the peopling of the Americas. Science Hagen, S. B. (2015) Y chromosome haplotype distribution of brown Advances, 2: e1501385. https://doi.org/10.1126/sciadv.1501385 bears (Ursus arctos) in Northern Europe provides insight into popula- McCarthy, T. M., Waits, L. P., & Mijiddorj, B. (2009). Status of the Gobi tion history and recovery. Molecular Ecology, 24, 6041–6060. bear in Mongolia as determined by noninvasive genetic methods. https://doi.org/10.1111/mec.13448 Ursus, 20,30–38. https://doi.org/10.2192/07GR013R.1 Servheen, C., Herrero, S., & Peyton, B. (1999). Bears: Status survey and McDevitt, A. D., Zub, K., Kawałko, A., Oliver, M. K., Herman, J. S., & conservation action plan. Gland: IUCN. Wojcik, J. M. (2012). Climate and refugial origin influence the mito- Sommer, R. S., & Benecke, N. (2005). The recolonization of Europe by chondrial lineage distribution of weasels (Mustela nivalis) in a phylo- brown bears Ursus arctos Linnaeus, 1758 after the Last Glacial Maxi- geographic suture zone. Biological Journal of the Linnean Society, 106, mum. Mammal Review, 35, 156–164. https://doi.org/10.1111/mam. 57–69. https://doi.org/10.1111/bij.2012.106.issue-1 2005.35.issue-2 Meiri, M., Lister, A. M., Collins, M. J., Tuross, N., Goebel, T., Blockley, S., Stamatakis, A. (2014). RAxML version 8: A tool for phylogenetic analysis ... Barnes, I. (2014) Faunal record identifies Bering isthmus condi- and post-analysis of large phylogenies. Bioinformatics, 30, 1312– tions as constraint to end-Pleistocene migration to the New World. 1313. https://doi.org/10.1093/bioinformatics/btu033 Proceedings of the Royal Society B: Biological Sciences, 281, 20132167. Stojak, J., McDevitt, A. D., Herman, J. S., Searle, J. B., & Wojcik, J. M. Molak, M., Lorenzen, E. D., Shapiro, B., & Ho, S. Y. W. (2013). Phyloge- (2015). Post-glacial colonization of eastern Europe from the Car- netic estimation of timescales using ancient DNA: The effects of pathian refugium: Evidence from mitochondrial DNA of the common temporal sampling scheme and uncertainty in sample ages. Molecular vole Microtus arvalis. Biological Journal of the Linnean Society, 115, Biology and Evolution, 30, 253–262. https://doi.org/10.1093/molbev/ 927–939. https://doi.org/10.1111/bij.12535 mss232 Svendsen, J. I., Heggen, H. P., Hufthammer, A. K., Mangeruda, J., Pav- Murtskhvaladze, M., Gavashelishvili, A., & Tarkhnishvili, D. (2010). Geo- lovc, P., & Roebroeksd, W. (2010). Geo-archaeological investigations graphic and genetic boundaries of brown bear (Ursus arctos) popula- of Palaeolithic sites along the Ural Mountains – On the northern tion in the Caucasus. Molecular Ecology, 19, 1829–1841. https://doi. presence of humans during the last Ice Age. Quaternary Science org/10.1111/mec.2010.19.issue-9 Reviews, 29, 3138–3156. https://doi.org/10.1016/j.quascirev.2010. Pavelkova Ri cankova, V., Robovsky, J., & Riegert, J. (2014). Ecological 06.043 structure of recent and Last Glacial mammalian faunas in Northern Swenson, J. E., Taberlet, P., & Bellemain, E. (2011). Genetics and conser- Eurasia: The case of Altai-Sayan refugium. PLoS ONE, 9, e85056. vation of European brown bears Ursus arctos. Mammal Review, 41, https://doi.org/10.1371/journal.pone.0085056 87–98. https://doi.org/10.1111/mam.2011.41.issue-2 Pavelkova Ri cankova, V., Robovsky, J., Riegert, J., & Zrzavy, J. (2015). Taberlet, P., & Bouvet, J. (1994). Mitochondrial DNA polymorphism, phy- Regional patterns of postglacial changes in the Palearctic mammalian logeography, and conservation genetics of the brown bear Ursus arc- diversity indicate retreat to Siberian steppes rather than extinction. tos in Europe. Proceedings of the Royal Society B: Biological Sciences, Scientific Reports, 5, 12682. 255, 195–200. https://doi.org/10.1098/rspb.1994.0028 Pitulko, V. V., Nikolsky, P. A., Girya, E. Y., Basilyan, A. E., Tumskoy, V. E., Taberlet, P., Fumagalli, L., Wust-Saucy, A. G., & Cosson, J. F. (1998). Koulakov, S. A., ... Anisimov, M. A. (2004) The Yana RHS site: Comparative phylogeography and postglacial colonization routes in Humans in the Arctic before the Last Glacial Maximum. Science, 303, Europe. Molecular Ecology, 7, 453–464. https://doi.org/10.1046/j. 52–56. https://doi.org/10.1126/science.1085219 1365-294x.1998.00289.x Rambaut, A., Lam, T. T., Carvalho, L. M., & Pybus, O. G. (2016) Exploring Talbot, S. L., & Shields, G. F. (1996). Phylogeography of brown bears the temporal structure of heterochronous sequences using TempEst (Ursus arctos) of Alaska and paraphyly within the Ursidae. Molecular (formerly Path-O-Gen). Virus Evolution, 2, vew007. https://doi.org/10. Phylogenetics and Evolution, 5, 477–494. https://doi.org/10.1006/mpe 1093/ve/vew007 v.1996.0044 Ramsden, C., Holmes, E. C., & Charleston, M. A. (2009). Hantavirus evo- Tamm, E., Kivisild, T., Reidla, M., Metspalu, M., Smith, D. G., Mulligan, C. lution in relation to its rodent and insectivore hosts: No evidence for J., ... Malhi, R. S. (2007). Beringian standstill and spread of native codivergence. Molecular Biology and Evolution, 26, 143–153. American founders. PLoS ONE, 2, e829. https://doi.org/10.1371/jour Ray, N., & Adams, J. M. (2001). A GIS-based vegetation map of the nal.pone.0000829 World at the Last Glacial maximum (25,000–15,000 BP). Internet Tammeleht, E., Remm, J., Korsten, M., Davison, J., Tumanov, I., Saveljev, Archaeology, 11, http://intarch.ac.uk/journal/issue11/rayadams_toc. A., ... Saarma, U. (2010) Genetic structure in large, continuous mam- htm) mal populations: The example of brown bears in northwestern Eura- Saarma, U., Ho, S. Y. W., Pybus, O. G., Kaljuste, M., Tumanov, I. L., Kojola, sia. Molecular Ecology, 19, 5359–5370. https://doi.org/10.1111/mec. I., ...Rokov,~ A. M. (2007) Mitogenetic structure of brown bears (Ursus 2010.19.issue-24 arctos L.) in northeastern Europe and a new time frame for the forma- To, T.-H., Jung, M., Lycett, S., & Gascuel, O. (2016). Fast dating using tion of European brown bear lineages. Molecular Ecology, 16, 401–413. least-squares criteria and algorithms. Systematic Biology, 65,82–97. https://doi.org/10.1111/j.1365-294X.2006.03130.x https://doi.org/10.1093/sysbio/syv068 Saarma, U., & Kojola, I. (2007). Matrilineal genetic structure of the brown Vulla, E., Hobson, K. A., Korsten, M., Leht, M., Martin, A.-J., Lind, A., & bear population in Finland. Ursus, 18,30–37. https://doi.org/10. Saarma, U. (2009). Carnivory is positively correlated with latitude 2192/1537-6176(2007)18[30:MGSOTB]2.0.CO;2 among omnivorous mammals: Evidence from brown bears, badgers Salomashkina, V. V., Kholodova, M. V., Tutenkov, O. Y., Moskvitina, N. S., and pine martens. Annales Zoologici Fennici, 46, 395–415. https://doi. & Erokhin, N. G. (2014). New data on the phylogeography and org/10.5735/086.046.0601 genetic diversity of the brown bear Ursus arctos Linnaeus, 1758 of Waits, L. P., Talbot, S. L., Ward, R. H., & Shields, G. F. (1998). Mitochon- Northeastern Eurasia (mtDNA control region polymorphism analysis). drial DNA phylogeography of the North American brown bear and ANIJALG ET AL. | 405

implications for conservation. Conservation Biology, 12, 408–417. SUPPORTING INFORMATION https://doi.org/10.1046/j.1523-1739.1998.96351.x Watson, T. (2017). News Feature: Is theory about peopling of the Ameri- Additional Supporting Information may be found online in the sup- cas a bridge too far? Proceedings of the National Academy of Sciences porting information tab for this article. of the United States of America, 114, 5554–5557. https://doi.org/10. 1073/pnas.1705966114 Willerslev, E., Davison, J., Moora, M., Zobel, M., Coissac, E., Edwards, M. E., ... Taberlet, P. (2014). Fifty thousand years of Arctic vegetation How to cite this article: Anijalg P, Ho SYW, Davison J, et al. and megafaunal diet. Nature, 506,47–51. https://doi.org/10. Large-scale migrations of brown bears in Eurasia and to 1038/nature12921 North America during the Late Pleistocene. J Biogeogr. Xie, W., Lewis, P. O., Fan, Y., Kuo, L., & Chen, M. H. (2011). Improving – marginal likelihood estimation for Bayesian phylogenetic model selec- 2018;45:394 405. https://doi.org/10.1111/jbi.13126 tion. Systematic Biology, 60, 150–160. https://doi.org/10.1093/sysb io/syq085

BIOSKETCH

Peeter Anijalg is interested in brown bear phylogeography and population genetics, and the impact of climate change on ancient and contemporary animal population processes. The remaining authors have broad interests in molecular ecology, phylogeography and biogeography, particularly relating to animals of the Late Pleistocene and Holocene.

Author contributions: U.S. and P.A conceived the original idea; all authors collected the data; P.A., S.Y.W.H., K.B., J.D. and U.S. analysed the data; P.A., S.Y.W.H., J.D. and U.S. wrote the first draft of the manuscript and all authors contributed to writing the final version.