Rich floras and faunas have been recovered from around the K/Pg boundary in the peri-Arctic regions (e.g. Akhmetiev & Beniamovski, 2009; Herman et al., 2009). To understand the biogeographical changes that took place, as well as the role of the Arctic as a geodispersal field, the temperature and the available light affecting the biota should be co-estimated (see the recent reviews by Spicer & Herman, 2010, and Eberle & Greenwood, 2012).
During the latest Cretaceous, angiosperms prevailed over ferns, conifers and cycadophytes in all peri-Arctic palaeofloras except the Alaskan (although pollen data from the Maastrichtian beds of the Prince Creek Formation reveal a higher diversity of angiosperms than that which is evident from the megafossils) (Spicer & Herman, 2010; and references therein). A strong differentiation had arisen between the north Alaskan and the north-eastern Russian regional palaeofloras since the Coniacian (and possibly as early as the late Turonian), suggesting limited gene flow and/or climatic differentiation that persisted into the Maastrichtian (Herman, 2007). Various sources of evidence (Nordt et al., 2003; Frank et al., 2005, and references therein; see Appendix S2) suggest, however, that the mid-Maastrichtian (CH3 highstand in Fig. 4) was a hyperthermal period [known globally as the mid-Maastrichtian event (MME)]. Hence, the observed limited genetic connection between the North American and north-eastern Russian floras is likely to result from geographical isolation caused by a transgressed Bering Strait (Fig. 7b,c). On the other hand, extended floral exchanges had taken place between the two areas by the early Palaeocene (Herman, 2007) (summarized below), suggesting a terrestrial communication (Fig. 7d); although the period corresponds to a lower warming event in the Kominz et al. (2008) eustatic sea-level curve (PH5 highstand in Fig. S1.3). This phenomenon could be explained by supposing that warmer periods result in higher eustatic sea levels leading to extended transgressive effects in low altitude areas.
Since the early Palaeocene, the regional palaeofloras of the peri-Arctic regions (such as northern Alaska, north-eastern Russia and Spitsbergen to Ellesmere Island) indicate relationships even at the species level, suggesting strong similarities in climatic regimes and genetic-terrestrial connections between the regions. The exclusive floristic relationships among the areas, if they are not due to collecting bias, especially support distinct dispersals via both the De Geer and the Bering routes (see Fig. 5a,b). Thus, Alaska and north-eastern Russia shared taxa such as Onoclea hesperia, Phragmites sp., Quereuxia sp. and Liriophyllum sp. during the early Palaeocene, and Tiliaephyllum brooksense and Archeampelos sp. during the late Palaeocene. These similarities suggest two distinct dispersals through Beringia, 65.5 Ma and c. 58 Ma, corresponding to the PH1 and PH5 sea-level highstands shown in Fig. 4 and Fig. S1.3 [for other evidence supporting warming periods during the highstands, see Quillévéré et al. (2008), Cramer et al. (2009) and Appendix S2 for 65.5 Ma, and Tripati et al. (2001) for c. 58 Ma].
On the other hand, north-eastern Russia and Spitsbergen shared taxa such as Coniopteris tschuktschorum, Elatocladus sp., Pseudolarix sp., Glyptostrobus nordenskioeldii and Taxodium sp.; and north-eastern Russia, Spitsbergen and North America shared taxa such as Cupressinocladus interruptus and Nordenskioldia borealis (for references see Fig. 5). Although some of these might be elements of a more ancient native floristic composition, others could suggest dispersals exclusively through the De Geer route. Indeed, Akhmetiev & Beniamovski (2009) (see also Akhmetiev, 2010) proposed that the desiccation of the epicontinental seas and adjacent parts of the Palaearctic Basin coincided with the Maastrichtian–Danian boundary, allowing westward migrations of thermophylic Tsagayan flora (Ginkgo, Pinaceae, Taxodiaceae, Trochodendroides, Platanaceae and Hamamelidaceae; i.e. the typical warm-temperate plants from the middle Amur Area of eastern Asia). Thus, at high latitudes, boreal humid deciduous flora migrated from the east to the west along the northern margins of the Siberian Platform and along the desiccated West Siberian Plate in the Danian, reaching the northern and middle Urals as well as Spitsbergen (via the De Geer route), where the same type flora has been recovered (at the Barentsburg locality; Akhmetiev & Beniamovski, 2009, and references therein). The occurrence of Palaeocarpinus joffrensis during the Palaeocene in North America and Spitsbergen (Storvola flora) is evidence for the exposure of the De Geer route (Golovneva, 2002; although in a different timeframe) that is, in fact, corroborated by the occurrence of this taxon in north-western China (Manchester & Shuang-Xing, 1996).
The discovery of the marsupial Maastrichtidelphys in the Maastrichtian of the Netherlands (Martin et al., 2005) is strong evidence for the existence of the De Geer route. Maastrichtidelphys exhibits similarities to early Maastrichtian North American herpetotheriids, providing definitive evidence of a high-latitude northern Atlantic dispersal route between North America and Europe during the latest Cretaceous (Martin et al., 2005). The occurrence of Maastrichtidelphys also demonstrates the existence of a junction between western Europe and Fennoscandia during the latest Cretaceous, which is further supported by vertebrate similarities between North America and Europe (Fig. 6). Nevertheless, this connection would have disappeared by the time the pantodonts occurred in both Asia and North America (early Palaeocene), which is concordant with the lack of pantodonts in the late Palaeocene (Thanetian) Cernaysian faunas of western Europe.
Recently, Tabuce et al. (2011) reported the recovery of a new mammal, Mondegodon eutrigonus, from the earliest Eocene of Silveirinha, Portugal. Mondegodon eutrigonus is considered, along with the early Palaeocene North American species Oxyclaenus cuspidatus, to be a morphological intermediate between two groups of ungulate-like mammals: the triisodontids and the mesonychians. Considering that the triisodontids are early to early–late Palaeocene North American taxa, Tabuce et al. (2011) proposed that Mondegodon probably belongs to a group that migrated from North America to Europe during the earliest Palaeocene and could thus represent a relict genus belonging to the Ante-Eocene European mammalian fauna. As such, the presence of Mondegodon in Europe constitutes more mammalian evidence of the exposure of the De Geer route.
Among the early Palaeocene mammalian faunas of North America and eastern Asia, the carnivorans, the mesonychids and the pantodonts are currently the only known taxa that are common to both areas and demonstrate eutherian similarities. Unequivocal carnivora are first known from the Palaeocene of North America. The two primitive families, Viverravidae and Miacidae, are often grouped as the paraphyletic Miacoidea (e.g. Rose, 2006). The oldest securely dated carnivoran, Ravenictis, comes from the earliest Palaeocene of Saskatchewan, Canada (Fox et al., 2010, and references therein). Ictidopappus (North America) and Pappictidops (China) are nearly as old and are known only from dentitions (Rose, 2006). They are regarded either as primitive viverravids or as basal carnivorans of uncertain affinity.
Three mesonychid genera (Yantanglestes, Hukoutherium, Dissacus) are known from the Shanghuan Asian land mammal ages (ALMAs) (Missiaen, 2011), whereas two (Dissacus, Ankalagon) are known from the Torrejonian (To2) North American land mammal ages (NALMAs) (Lofgren et al., 2004). The clade represented by the genus Dissacus probably dispersed more than once between the continents.
The pantodonts possess traits common to the mammalian faunas of Asia, North America and South America (McKenna & Bell, 1997; Lucas, 1998). Among the early pantodonts are the primitive Alcidedorbignya from the Tiupampa in south-central Bolivia (de Muizon & Marshall, 1992, and references therein), the Chinese genera Bemalambda and Hypsilolambda (e.g. Missiaen, 2011), and the North American Pantolamda and Titanoides (Lofgren et al., 2004), all of which are from the early Palaeocene. Recently, large footprints of a Titanoides-like pantodont were discovered in the Todalen Member coal layers, providing the earliest evidence of a large mammal on Svalbard and the northernmost example of the presence of a land mammal from the Palaeocene (Lüthje et al., 2010). The presence of a Palaeocene pantodont on Svalbard is consistent with the concept of the De Geer route.
Primitive ungulates such as the rabbit-sized arctostylopids and the uintathere Prodinoceras are first known from the Nongshanian and the Gashatan ALMAs, the former from both and the latter from the Gashatan only (e.g. Missiaen, 2011). Since the Tiffanian (Ti5, c. 58 Ma), both clades are also represented in North America by the genera Arctostylops and Prodinoceras (Lofgren et al., 2004). These ungulates are proposed to have dispersed between Asia and North America via Beringia (e.g. Missiaen, 2011). This view is supported by evidence that the Ti5 age coincides with observed floristic similarities and the PH5 sea-level highstand suggested by the Kominz et al. (2008) eustatic curve. Therefore, all of the first occurrences within the Ti5 times of the North American taxa related to Asian forms are likely to have migrated to North America via Bering route 2. According to the North American mammalian biostratigraphy of Lofgren et al. (2004), such Ti5 migration events could include the first occurrences of the viverravid genera Viverravus and Didymictis; the anagalid Mingotherium, which has been related to the endemic Asian family Pseudictopidae (Rose, 2006); the metacheiromyid Propalaeanodon, which could be related to Asian palaeanodonts (Rose, 2006); the oxyaenid genera Oxyaena and Palaeonictis (oxyaenids are considered to be immigrants from Asia; Rose, 2006); as well as the carpolestid plesiadapiforms (Smith et al., 2004).
The largest arctocyonid was the dog-to-bear-sized Arctocyon, which was present in the late Palaeocene of both Europe and North America (in the Thanetian and the Torrejonian–Tiffanian, respectively) (Rose, 2006). The first occurrence of Arctocyon in Europe is dated to 58.5 Ma (Hooker & Collinson, 2012). This age is well after the exposure of the De Geer route, as proposed in the current study, and slightly before the exposure of Bering route 2 (c. 58 Ma). Given the absence of Arctocyon from Asia, the unexpected occurrence of Arctocyon is likely to be a ‘sweepstake’ dispersal through the early and incomplete formation of the Thulean route near the end of the first magmatic phase of the North Atlantic Igneous Province (NAIP) c. 58 Ma. In Fig. S1.3, the occurrence of the North American Arctocyon in Europe correlates with sea-level lowstand PL4 and the first unconformity of the Faeroe-Shetland Basin (as considered in Appendix S1), implying a restriction of the marine barrier.
The geological evidence hypothesized in Appendix S1 to document the first full exposure of the Thulean route (LGR81) c. 57 Ma (Fig. 3d, Fig. S1.3) is in agreement with additional mammalian evidence. The recently described plesiadapid Platychoerops antiquus from France may be the result of a dispersal of the North American Plesiadapis cookei in the late Thanetian (Boyer et al., 2012). A similar dispersal could account for the mesonychid Dissacus, known from the Torrejonian of North America and the Gashatan of Asia (Missiaen & Smith, 2008) as well as from the Cernaysian of Europe (first occurrence c. 57 Ma; Hooker & Collinson, 2012). The direction of these dispersals was probably from North America to Europe via the Thulean route and from Europe to Asia via a regressed Turgai Strait (Fig. S1.3). The likelihood of this scenario depends on the age of the first occurrence of Dissacus in Asia. If Dissacus appeared > 57 Ma, then it could have dispersed via Bering route 2 c. 58 Ma [the possible occurrence of the same genus in the Shanghuan ALMA (Missiaen, 2011) is not considered here]. The same course (i.e. via the Turgai Strait), but in the opposite direction could account for the dispersal of the Rodentia. Previously described only from the late Palaeocene of Asia and North America, they are now also known from comparable (late Thanetian) beds in Europe (Smith et al., 2010b). Thus, the Rodentia probably occurred first in Asia and then reached North America via Europe.
According to Lofgren et al. (2004), the first occurrence of Viverravus in North America is at Ti5, suggesting dispersal via Bering route 2 c. 58 Ma (PH5 sea-level highstand). Comparable morphotypes to Viverravus are also known from the Silveirinha Formation in Portugal (Antunes et al., 1997; Estravis, 2000) and are probably of latest Palaeocene age (Pais et al., 2012, and references therein). Therefore, the genus Viverravus may have dispersed to Europe either from Asia via the Turgai Strait or from North America via the Thulean route.
The first occurrence of Coryphodon in North America at the onset of Cf1 (Clarkforkian) (Lofgren et al., 2004), and not at Ti5 (Tiffanian) (i.e. at the time of favour climatic conditions for dispersal via the Bering route 2), suggests a later dispersal via the Thulean route in the late Palaeocene, although this inference cannot be confirmed by the current fossil record because in Europe Coryphodon is known only from the Palaeocene–Eocene boundary (PE I; Hooker & Collinson, 2012). Coryphodon teeth are among the most common mammalian fossils in the Eureka Sound Group of Ellesmere Island, where the mesonychid genus Pachyaena was also recovered (Eberle & McKenna, 2002). Pachyaena is known from the late Palaeocene of Asia (Meng et al., 2005) and the earliest Eocene of Europe (PE II; Hooker & Collinson, 2012). Its slightly earlier presence in North America (Wa0 versus PE II) could be interpreted as dispersal via a possible earliest Eocene exposure of Beringia (not considered here), rather than from Europe via a combination of the Turgai Strait and the Thulean route.