Russian scholars (e.g., Kharzinova et al. 2018; Mizin et al. 2018; Rozhkov et al. 2020; Vasilchenko et al. 2020) recognise 4–8 subspecies within Russia (Fig. 3): those mentioned above plus angustirostris. Of these, fennicus, valentinae, angustrostris, and phylarchus are forest reindeer and are larger, longer-legged, and darker and have shorter, heavier, and more branched antlers than tundra reindeer (Baskin 1986; Rozhkov et al. 2020). Although Wilson and Reeder (2005), followed by Wilson and Mittermeier (2011), chose R. t. buskensis Millais, 1915, as a senior synonym for R. t. valentinae, Millais (1915a) is not a valid taxonomic authority (see Discussion).

Because of Banfield’s (1961) lumping, Western scholars have often not distinguished between true R. t. tarandus of the western European mountains, and the far more widespread R. t. sibiricus. For this reason, papers on genetic diversity must be read carefully to determine the provenance of the specimens.

Although Eurasian tundra reindeer and North American barren-ground caribou are distinguishable by different allele frequencies at several loci, they have low genetic differentiation (Cronin et al. 2005). They have diagnostically different pelage patterns and other morphological differences (see Geist 1998 for descriptions).

The "mountain reindeer” of Norway (and formerly Sweden and Finland) are tundra reindeer (R. t. tarandus) that have adapted to high-elevation alpine tundra with snow characteristics similar to Arctic tundra: hard-packed, shallow snow that they can paw through to reach terrestrial lichens. The haplotype composition of reindeer from southern Norway is similar to, but "substantially genetically different from” that of the tundra reindeer from western Russia, R. t. sibiricus (von Schreber, 1784) (Baranova et al. 2012).

Rangifer t. sibiricus includes 19 herds, named for their calving grounds, from Arkhangelsk in European Russia to Chukotka, Siberia (Mizin 2018; Rozhkov et al. 2020). Although domestic reindeer descended from tundra types, there is a "clear genetic differentiation between domestic and wild reindeer populations” with ~ 6% introgression by wild reindeer into domestic clades and none the other way (Kharzinova et al. 2018; Rozhkov et al. 2020). There is little genetic difference among wild tundra populations of R. t. sibiricus in Taymyr, northern Yakutia, and Chukotka (Kharzinova et al. 2018).

Based on mtDNA, wild reindeer in Genhe, north of the Greater Khingan Mountains in Heiliongjiang, China (Temperate Coniferous Forest zone, Fig. 3), are of Beringian-Eurasian lineage semi-domesticated by the Ewenki people that lost their migratory habits, not forest reindeer as previously supposed (Wang et al. 2019; Ju et al. 2020).

Despite Lydekker’s (1915) attempt to bring it under R. tarandus, Sokolov (1937, 1963) insisted that its skull shape, especially the rostrum, and the dentition, were different enough to maintain R. platyrhynchus Lönnberg, 1909 as a species. Svalbard (and the extinct east Greenland and Peary’s) caribou derived from ancient Beringian-Eurasian pre-glacial populations, based on shared mtDNA haplotypes, but evolved in separate refugia during the LGM (Flagstad and Røed 2003; Kvie et al. 2016). Svalbard reindeer (with West Greenland caribou; see below) are the most genetically distinct of all Rangifer subspecies (average genetic differentiation [Fixation Index, F_ST] 41% to all other reindeer and caribou); they are not, however, closely related to each other, with a 69% genetic differentiation between them (Yannic et al. 2013). The large genetic differentiation qualifies Svalbard reindeer as R. platyrhynchus Lönnberg, 1909 (Miller 1912a).

Based on mtDNA control region sequences, reindeer of the High Arctic archipelagos of Franz Josef Land and Novaya Zemlya, R. t. pearsoni Lydekker, 1902, descended from wild tundra reindeer from the Eurasian mainland after the LGM, but before humans could have brought domesticated reindeer (Kvie et al. 2016). Reindeer occupied Franz Josef Land briefly, having migrated there after the Holocene climatic optimum (~ 6,000–4,500 years ago) when the climate again became colder and the sea‐ice more persistent, and became extinct historically (Mizin et al. 2018). Novaya Zemlya reindeer occupied the island only 7,000–5,000 years ago and were described by their distinctive phenotypical appearance (Lydekker 1902); however, today’s population is mostly feral domestic reindeer, the endemic R. t. pearsoni having either not survived or the few remaining individuals mixed with domestic reindeer (Mizin et al. 2018 and references therein).

The Okhotsk or Kamchatka reindeer, R. t. phylarchus Hollister, 1912, has pelage patterns and antler formation more like Canadian barren-ground caribou than other Eurasian subspecies, prompting Geist (1998) to conclude that, "This is no reindeer, but a caribou.” It probably dispersed from Beringia in the late Pleistocene in a "second radiation into Siberia”, after Canadian and Eurasian forms had evolved distinctive patterns and adaptations (Geist 1998). Rozhkov et al. (2020) showed that wild reindeer from Kamchatka cluster separately from those living west of the Sea of Okhotsk, which are indistinguishable genetically from the Jano-Indigirka, East-Siberian taiga, and Chukotka populations of R. t. sibiricus (von Schreber, 1784). The range of Kamchatka reindeer therefore should be restricted to the Kamchatka Peninsula.

Finnish or European forest reindeer, R. t. fennicus Lönnberg, 1909, was described from Finnish Lapland (Fig. 3). Ellerman and Morrison-Scott (1951) synonymised it with the Kerelian forest reindeer R. t. silvicola Hilzheimer, 1936 and Siberian forest forms R. t. transuralensis Hilzheimer, 1936, R. t. dichotomus Hilzheimer 1936, and R. t. angustirostris Flerov, 1932 as junior synonyms. That it is considerably larger than R. t. tarandus "can hardly be due to nutritional factors alone”; it also has:

"significantly longer legs...[that] are an important adaptation to taiga conditions, where the snow cover is usually deep and soft. The mountain types [R. t. tarandus in Norway] have evolved in areas with hard-packed tundra snow, and consequently the semi-domestic reindeer have difficulty surviving in coniferous forests, especially in winters with deep, soft snow (Nieminen and Helle 1980).”

Rangifer t. fennicus has statistically significant cranial differences from tundra reindeer, particularly its arched nasal bones (flattened in tundra reindeer: Banfield 1961). Its antlers are "…reminiscent of barren-ground antlers, but with oval beams and a bez tine set well above the brow tine…shaped like tundra-type antlers but more massive and show some palmation” (Geist 1998). Its pelage pattern is similar (see Geist 1998 for descriptions).

Rangifer t. fennicus evolved in isolation from the tundra type in a separate western European refugium and adapted to forest environments; it shares no mtDNA haplotypes with any North American caribou (Flagstad and Røed 2003; Røed 2005).

Between wild tundra and taiga reindeer (subspecies not stated; presumably R. t. fennicus sensu lato) pairwise F_ST values, using a "genome-wide bovine SNP genotyping array”, averaged 3.8%–9.4%, "consistent with their morphological and ecological differences” (Kharzinova et al. 2018). Weldenegodguad et al. (2020), based on microsatellites, also found that Finnish reindeer clustered separately from all other ecotypes (Eurasia tundra, Alaska, and Svalbard). Genetic distances (see Suppl. material 2: Genetic distance) and differences in morphology and ecology between fennicus and tarandus suggest that the former should be returned to full species status, with subspecies R. fennicus fennicus, R. f. valentinae, and possibly R. f. angustirostris.

Rangifer t. angustirostris, the East Siberian forest reindeer, currently numbers ~ 1,000 animals, distributed east of Lake Baikal (Mizin 2018). Its status, whether more allied to fennicus than to sibiricus, is best "left in doubt until data on its genetics become available” (Rozhkov et al. 2020).

Croitor (2010) hypothesised that R. t. fennicus evolved from Cervus guettardi Desmarest, 1822, a reindeer that adapted to forest habitats in western Europe as forests expanded during an interglacial period before the LGM (the Würmian or Weichsel glaciation); guettardi was later replaced by R. constantini, a more evolved tundra form (cf. Baranova et al. 2016), in a second immigration 19,000–20,000 years ago when the LGM turned its forest habitats into tundra, while fennicus survived in isolation in south-western Europe. If correct, fennicus does not share a common ancestor with R. tarandus and cannot be conspecific. Its name would be Rangifer fennicus Lönnberg (Miller 1912a).

Wild reindeer from Murmansk/Kola Peninsula are forest reindeer, R. t. fennicus, sharing a clade with those from Karelia and Arkhangelsk; these share two haplotypes with domestic reindeer from the same regions, but show only a low incidence of hybridisation, indicating ancient introgression (Baranova et al. 2016; Korolev et al. 2017; Vasilchenko et al. 2020).

The Altai-Sayan forest reindeer (Rangifer tarandus valentinae Flerov, 1933) is a montane form whose ecology parallels that of British Columbia’s mountain caribou (see below). It migrates altitudinally in dense coniferous forests at elevations of 400–1,500 m, where snow cover is 130–250 cm, and forages arboreal lichens in winter (Baskin 1986). Its mating system also is similar to that of British Columbia mountain caribou: males guard 3–5 females during rutting season and calving is dispersed in alpine habitats (Shaposhnikov 1955, cited by Sobansky 1976). It has a unique mitochondrial genome and shows no signs of introgression of domestic reindeer mtDNA (Vasilchenko et al. 2020). Its genetic differentiation (using a genome-wide genotyping array to compare single nucleotide polymorphism, SNP, markers) from three populations of putative R. t. sibiricus is F_ST = 3.1%–3.5% (Kharzinova et al. 2018). See Geist (1998) for descriptions of pelage patterns.

Interestingly, the two forest forms, fennicus and valentinae, cluster together as sister clades, based on mtDNA haplotypes, even though separated by 3,800 km and with the East European Taiga population of sibiricus between them; and these two form a sister clade with a Siberian taiga population of sibiricus to the exclusion of tarandus, pearsoni, and phylarchus (Rozhkov et al. 2020). This qualifies them as R. fennicus valentinae. Davydov et al. (2007) also united valentinae with fennicus as closely-related subspecies that clustered apart from tundra and Arctic island forms in Eurasia and North America.

Early genetic analyses showed two major lineages of caribou in North America: migratory barren-ground caribou, whose ancestors survived the LGM in Beringia, that calve on the tundra and migrate in winter to boreal forest; and a non-migratory, exclusively forest clade whose ancestors persisted south of the ice-sheets that covered northern North America and the western cordillera (Courtois et al. 2003; Flagstad and Røed 2003; Zittlau 2004; Cronin et al. 2005). COSEWIC (2011) and others refer to these as the BEL (Beringian-Eurasian) and NAL (North American) lineages, respectively.

Currently recognised Canadian BEL barren-ground caribou subspecies are R. t. groenlandicus sensu lato of the mainland tundra, R. t. caboti of Labrador, R. t. osborni of the northern cordillera, R. t. pearyi of the High Arctic, and the extinct insular R. t. dawsoni (Wilson and Mittermeier 2011). However, Banfield (1961) erred, both in failing to recognise the species-level separation of R. tarandus from R. arcticus, and in assigning the subspecies name groenlandicus to mainland barren-ground caribou, as discussed below. Its proper name is R. arcticus Richardson, 1829 (Allen 1942).

All three western montane ecotypes (Osborn’s caribou, Rocky Mountain caribou and Selkirk caribou: Fig. 2) are of BEL ancestry, but are deeply divergent genetically and ecologically, having split from barren-ground caribou some 60,550 years ago in the Illinois-Wisconsin interglacial; each is a different combination of separate BEL lineages (Klütsch et al. 2012; Polfus et al. 2017; Taylor et al. 2021).

The "southern group of the Southern Mountain population of Woodland caribou”, R. tarandus caribou (cf. COSEWIC 2014) was originally described as the Mountain or Selkirk caribou, R. montanus Seton-Thompson, 1899; the Central Mountain population was Rocky Mountain caribou, R. fortidens Hollister, 1912; and the Northern Mountain population was Osborn’s caribou, R. osborni Allen, 1902. These scientists distinguished the three mountain types based on quantitative differences in dentition, skeletal and antler measurements, pelage colour and size, as well as ecology.

Anderson (1946) concurred with Jacobi (1931) in retaining the subspecies designations of all three western montane ecotypes under Arctic caribou: R. arcticus osborni, R. arcticus montanus, and R. arcticus fortidens.

When Ellerman and Morrison-Scott (1951) revised Rangifer into a single species, the Eurasian name, R. tarandus taking priority, Selkirk caribou became R. t. montanus, and Osborn’s caribou, R. t. osborni (McTaggart Cowan and Guiguet 1956). Significantly, although Banfield (1961) acknowledged that his measurements showed Osborn’s and Selkirk caribou as morphologically separate from each other and from barren-ground caribou and woodland caribou, he still lumped them with R. t. caribou Gmelin, 1788. Nagorsen (1990), then Curator of Mammals at the Royal British Columbia Museum, objected: "...these two morphs [Mountain and Osborn’s] exhibit some differences in size, antler morphology and pelage colour… a modern study of geographic variation … is needed to resolve the systematics of woodland-mountain caribou”.

Geist (1991) maintained the separation of montanus from osborni on the basis of size, pelage patters and colour, montanus being smaller and darker. Osborn’s caribou, currently recognised as R. t. osborni (Wilson and Mittermeier 2011), therefore reverts to R. a. osborni Jacobi (1931).

Serrouya et al. (2012), Harding et al. (2020) and others called the Selkirk caribou "deep-snow mountain caribou”, because, uniquely, they winter high on the mountains where they walk on top of a 2–5 m deep snowpack to forage arboreal lichens. The name, Rangifer arcticus montanus Seton-Thompson, 1899 (Jacobi 1931) is available; or Rangifer montanus Seton-Thompson, 1899, as Murie (1935) insisted and as its genetic distance from others (see above and Suppl. material 2: Genetic distance) suggests.

Rocky Mountain caribou, or the Central Mountain population DU8 per COSEWIC (2011), occupy the east slope of the Rocky Mountains (Fig. 2) where the continental climate results in light, shallow snow in which they forage terrestrial lichens in winter. They average 55 km horizontal migration to forested winter ranges, a little less than Osborn’s caribou and far more than "sedentary” boreal woodland caribou (Theoret et al. 2022). They are mountain caribou that have hybridised in ancient times with woodland caribou, with which they share haplotypes (McDevitt et al. 2009; Taylor et al. 2021). The name R. a. fortidens Hollister, 1912 (Jacobi 1931) is available and appropriate.

A caribou antler from Haida Gwaii, British Columbia was dated to ~ 43,200 years BP in the mid-Wisconsin Olympia Interglacial (Mathewes et al. 2019). More recent (4,000–6,000 BP) bones of the extinct, insular Dawson caribou, originally R. dawsoni Seton-Thompson, 1900, were similar to barren-ground caribou but smaller, evidence of island dwarfism (Byun et al. 2002 and references therein). Evidently, they evolved in a coastal refugium after the LGM when rising sea levels isolated them. Byun et al. (2002), using molecular and ancient DNA techniques, were able to sequence a short fragment of the mtDNA from remains of Dawson caribou. Their results allied them phylogenetically with the caribou on the adjacent mainland (which at the time were thought to be R. t. caribou, but which are now known to be BEL lineage R. a. osborni) and a little less closely to Alaska barren-ground caribou (see Suppl. material 2: Genetic distance).

In Alaska, of 13–32 caribou herds that have been recognised, including four that overlap with Yukon (Hemming 1971; Valkenburg 1998), four (Porcupine, Central Arctic, Teshekpuk, and Western Arctic) are small, phenotypically barren-ground caribou with long migrations and aggregated, tundra calving grounds (Prichard et al. 2020), while a fifth (Steese-Fortymile) is intermediate in form and behaviour (Hemming 1971; Gauthier and Farnell 1986; Valkenburg 1998; Mager et al. 2014).

Allen (1902) described Grant’s caribou, R. granti Allen, 1902, of the Alaska Peninsula and archipelago (Fig. 2), noting that it was "not closely related to R. stonei of the Kenai Peninsula, from which it differs not only in its very much smaller size, but in important cranial characters and in [pale] coloration.” It remained a species or subspecies (see Suppl. material 1: Synonymy) until Banfield (1961) erroneously brought all other Alaska caribou under it as junior synonyms, thus expanding its range to the whole state and northern Yukon. Youngman (1975) began a trend to replace granti with R. t. groenlandicus sensu lato. Because Geist (1998) could find no morphological features to distinguish Alaskan from Canadian barren-ground caribou, granti was not accepted by Wilson and Reeder (2005) and Wilson and Mittermeier (2011). Caribou geneticists agree that they are barely distinguishable (e.g., Cronin et al. 2005; Yannic et al. 2018). As originally described, however, granti survives (see below).

Murie (1935) brought Rangifer excelsifrons Hollister, 1912, Rangifer mcguirei Figgins, 1919, and Rangifer ogilviensis Millais, 1915, under R. stonei Allen, 1901 (Fig. 2). Stone’s caribou was said to range from the Kenai and Alaska Peninsulas to western Yukon "and more sparingly to the eastward” (Murie 1935): a large, dark caribou with a well-developed white fringe on the throat and "antlers large and rangy, of the arcticus type, but heavier”. The former R. ogilviensis (Millais, 1915b) is the Porcupine Herd (named for a river that flows from Yukon into Alaska) of barren-ground caribou that winters mainly in the Ogilvie Mountains, Yukon, and calves on the Alaska-Yukon coastal plain. Their migratory, rutting and calving-aggregation behaviours and small size (consistently smaller than mountain caribou to the south and west: Gauthier and Farnell 1986) reveal their barren-ground identity.

In Alaska, some two dozen herds are genetically, morphologically (larger and darker than barren-ground caribou: Gauthier and Farnell 1986) and ecologically similar to the western Canada montane forms (Mager et al. 2014 sampled 20 of the 26 currently recognised herds). Nevertheless, they clustered clearly into two groups at K = 2, one of which encloses the Alaska Peninsula type locality of R. granti (Fig. 2). Colson et al. (2014) found a lack of dispersal or introgression from adjacent ecotypes, suggesting reproductive isolation of the Alaska Peninsula/archipelago cluster. Yannic et al. (2018) confirmed the genetic distinctiveness of this ecotype, which had been previously found to differ morphologically as well (cf. Murie 1935; see Suppl. material 1: Synonymy). Thus, R. a. granti is rediscovered, its range restricted (as originally: Allen 1902) to the Alaska Peninsula and archipelago.

At K = 4, six "mainland” (i.e., not peninsula/island) herds, all geographically small, isolated mountain herds, "appeared relatively discrete with > 0.50 population assignment to one cluster, rather than several [and] had high pairwise differentiation” (Mager et al. 2014). One of these, the Chisana herd, which spans the Alaska-Yukon border, contains the type locality (Fig. 2) of Rangifer mcguirei Figgins, 1919. Figgins (1919) found differences, which he thought sufficient to name a new species, in the pelage, cranial and dental features, and antler formation of six specimens of Chisana caribou, compared to the same morphological characters in osborni and stonei. Murie (1935), followed by Anderson (1946), synonymised mcguirei with stonei because "No part of the original description would distinguish the type specimen from R. arcticus stonei. Furthermore, the type locality lies squarely in the path of migration of the large herd of stonei, the principal herd of Alaska-Yukon caribou, at a point where hundreds of thousands pass through each year during the rutting season.”

This reasoning is flawed because none of these mountain herds migrate long distances: rather, each migrates altitudinally if at all (most winter in alpine tundra where wind clears snow from the terrestrial lichens) and maintains separate alpine rutting and calving areas. Murie (1935) no doubt had observed the Steese-Fortymile herd on migration. Nevertheless, Osgood (1909) and Murie (1935) were prescient in bringing all of these under R. arcticus.

Likewise, 16 southern Yukon and northern British Columbia herds, 15 of them currently identified as R. t. osborni, clustered into four groups based on microsatellite DNA analysis and three based on mtDNA (Kuhn et al. 2010). Haplotypes of the migratory Steese-Fortymile herd (the others are sedentary), were spread throughout the others, suggesting it as ancestral to all with perhaps occasional introgression.

The clustering pattern described above (Kuhn et al. 2010; Mager et al. 2014) argues for restoring the subspecies name, R. a. stonei Allen 1901 (Murie 1935), to Chisana and all other interior Alaska mountain caribou that cluster together and apart from osborni, granti sensu stricto, and Alaskan barren-ground caribou, including Steese-Fortymile, as subspecies of R. arcticus. This is another case of the pre-Banfield taxonomy being confirmed by genetic data.

Introductions of R. t. tarandus sensu lato and R. t. sibiricus into Alaska and thence to Nunavut were detailed by Anderson (1946). Some interbreeding between reindeer and wild caribou in Alaska has been documented, but with very little introgression in either direction, probably because of low fitness of hybrid animals in the wild; relatively little crossbreeding has been observed when the two have been in captivity together (Cronin et al. 2006).

Banfield (1961) unjustifiably renamed R. arcticus as R. t. groenlandicus (see below). It has seven recognised herds on mainland Canada, defined by calving grounds, and extends into Alaska. Barren-ground caribou are smaller and paler than woodland caribou, but have longer, thinner antlers that, in males, sweep back, up and forward, main beams reaching > 135 cm with multiple tines at the top, often palmated (see Geist 1998 for diagnostic features). Its beams are round in cross-section vs. flattened in woodland caribou.

Anderson (1913) described the migration of the Dolphin and Union herd, DU2 (Fig. 2), across Dolphin and Union Straits from Victoria Island to the mainland and back. He also briefly described its diagnostic pelage pattern and cranial morphology (e.g., "...The heads of these Caribou appeared to be much shorter than those of the Great Bear Lake Caribou, with a noticeable fullness or convexity between forehead and nose”). He intended to name a new form of caribou and to select a type specimen from among the 84 he collected in 1911 and sent to the American Museum of Natural History; any of mature males AMNH 34431, 34432, and 34435 would be a suitable holotype. He also sent 24 to what is now the Canadian Museum of Nature, but never completed a formal report of his second (1913–1916) expedition (Anderson 1917). Thomas and Everson (1982) later confirmed its unique skeletal features quantitatively. It was long thought to be either a race of Peary caribou or a hybrid between Peary and barren-ground caribou (e.g., Manning 1960) until genetic and other data showed it to be a distinct race of barren-ground caribou (Zittlau 2003; COSEWIC 2004, 2017). Since it was never formally described, there is no available subspecies name for this ecotype.

Peary’s caribou, R. t. pearyi, of the Arctic Archipelago except for Baffin Island (Fig. 2), is "clearly most genetically similar to the Canadian barren-ground caribou (R. t. groenlandicus) from North West Territories … suggesting common origin of the ancestors of these populations” (Røed 2005). Peary caribou diverged from barren-ground caribou 96,000–185,000 BP and evolved in isolation through two glacial cycles (Klütsch et al. 2017). COSEWIC (2004) restricted its distribution to the High Arctic islands and the Boothia Peninsula, except for most of Victoria Island, based on Harding (2003) and Zittlau (2003). A BEL lineage, it is DU1 (COSEWIC 2011).

Baffin Island caribou comprise insular populations that are geographically and genetically disjunct from both mainland barren-ground and Peary caribou (Fig. 2). It differs from the mainland barren-ground caribou in lacking large-scale migrations and with calving being dispersed rather than aggregated (Jenkins et al. 2012). Its genetic differentiation (pairwise F_ST based on 16 microsatellite loci) "is evidence of evolutionary significance and points to Baffin Island caribou as a candidate for consideration as a DU” (Jenkins et al. 2018).The most common allele in Baffin Island caribou is absent or occurs in very low frequencies in other barren-ground caribou populations including the nearby Beverly herd on the adjacent mainland; likewise, the Beverly herd has eight alleles that are absent from the Baffin Island samples, indicating "a large genetic distance” between them (gel electrophoresis: Røed et al. 1991). Baffin Island caribou share one haplotype (C10) with those in Labrador and two each with Dolphin and Union and Bluenose herds of barren-ground caribou (Cronin et al. 2005). The genetic distances (see Suppl. material 2: Genetic distance) and other data suggest at least a subspecies. There is no available name for a Baffin subspecies.

Small, pearyi-sized caribou occupied the ice-free parts of Greenland in the Illinoian-Wisconsin interglacial and through the LGM and early Holocene (Meldgaard 1986). Degerbøl (1957) described R. t. eogroenlandicus, which became extinct ~ 1900, from a relict enclave in north-eastern Greenland (Fig. 2). However, Bennike (1988), comparing bones and noting that Peary caribou have been documented crossing Nares Strait to Greenland, doubted that pearyi and eogroenlandicus were subspecifically distinct. That Peary caribou shared certain mtDNA haplotypes and morphological similarities with it (Kvie et al. 2016) casts further doubt on the validity of R. t. eogroenlandicus.

The larger West Greenland caribou is problematic. It is darker than typical arcticus and much darker than pearyi, resembling woodland caribou in its dark-brown body, with neck and ventral area much lighter (Allen 1908). Allen (1908) gives Greenland caribou average condylobasal length and upper maxillary tooth row metrics, both greater than in mainland arcticus and considerably more than in the "little pearyi”. He also notes that antlers of Greenland caribou adult males, although within the range of arcticus in total length, are "slenderer, less palmated, and more recurved”. Historic and archaeological records show that barren-ground-sized caribou immigrated to West Greenland, possibly from Baffin Island via the Davis Strait, in the middle Holocene (Meldgaard 1986). However, a reconstruction of glacial retreat and caribou advance (Yannic et al. 2013) shows colonisation by NAL lineage caribou more likely.

Greenland caribou, with Svalbard caribou, are the most geographically and genetically isolated ecotypes among all extant caribou (average fixation index 41%: Yannic et al. 2013) based on 16 microsatellite loci. They share low relatedness values with all Canadian caribou (Solmundson et al. in press).

The (West) Greenland caribou is neither of the BEL lineage, from which descend all Eurasian and Canadian tundra reindeer and caribou, nor the NAL lineage of woodland caribou: it clusters outside of the BEL cluster, as do Svalbard reindeer (Yannic et al. 2013). Yannic et al. (2018) were unable to include it in their hierarchical analysis "because of their high genetic differentiation”. It is best left as Linnaeus (1767) originally classified it: Rangifer groenlandicus L., 1767 (Miller 1912a).

Polfus et al. (2017), using nuclear and mtDNA, discovered another distinct "woodland” clade in the Northwest Territories between the Mackenzie River and Great Bear Lake that descends, not from the NAL lineage as other woodland caribou, but from the BEL lineage. They quoted Geist (2007), who, using pelage and antler characteristics and taxonomic inferences, suggested that the mountain and boreal "woodland” caribou north of 60° latitude were more likely "splinter populations of barren ground caribou, which have adapted to a more sessile life-style, increased in body size, and assumed some ‘woodland mannerisms’”. Since the best-supported model of Polfus et al. (2017) shows descent from R. a. osborni, it should be considered an ecotype of Osborn’s caribou.

Gmelin (1788), editing the 13^th edition of Linnaeus’ ‘Systema Naturæ’, carried forward Brisson’s (1756) "Le Karibou” as a subspecies, Cervus tarandus γ caribou. Gmelin (1788) did not name a type locality. Miller 1912b and Lydekker (1915) both gave it as "Eastern Canada” without attribution. Banfield (1963) later designated a neotype and neotype locality at Quebec (City), Province of Quebec, Canada. In Fig. 2, I placed the type locality symbol in the middle of the Province of Quebec.

Historically, most Rangifer taxonomists (e.g., Baird 1859; Ross 1861; Allen 1896; Lydekker 1898; Miller 1912b) thought the differences from barren-ground caribou warranted species status, R. caribou, based on the larger size, darker colour, different antlers, and sedentary habits (see Suppl. material 1: Synonymy). Lydekker (1915) described woodland caribou antlers as "...stout, flattened, much palmated, and not of excessive length, one of the brow-tines being much expanded, while the other is simple; the bez-tine is also more palmated than in the Scandinavian reindeer.”

Genetic and morphological analyses (e.g., Klütsch et al. 2012; Horn et al. 2018; Yannic et al. 2018) have confirmed the woodland caribou’s distinctiveness and divergence from other ecotypes ~ 357,000 BP (Horn et al. 2018). These differences and its genetic distance from other ecotypes (see Suppl. material 2: Genetic distance) warrant restoration of Rangifer caribou Gmelin, 1788 (Baird 1859).

Three haplogroups of woodland caribou are evidence of isolation in three refugia (in the Rocky Mountains, east of the Mississippi, and the Appalachian Mountains) during the LGM, giving rise to two clades of boreal woodland caribou east and west of a "suture zone” in Manitoba (Klütsch et al. 2012). A fourth, coastal refuge was later identified (Wilkerson et al. 2018). Mid-continent clades have a few barren-ground haplotypes, arising after the LGM when the latter rutted far enough south to encounter the former; however, barren-ground caribou do not have woodland caribou haplotypes (McQuade Smith 2009; Yannic et al. 2018).

COSEWIC (2011), based mainly on molecular data, divided Banfield’s (1961) woodland caribou, R. t. caribou, into five Designated Units (Fig. 2): Eastern Migratory DU4, Newfoundland DU5, Boreal DU6, Torngat DU10, and Atlantic-Gaspésie DU11, in addition to the three western montane ecotypes discussed above.

Richardson (1829), without specifying a type locality, had described the "western woodland caribou”, R. t. sylvestris (Richardson, 1829). He gave its range as west of Hudson’s Bay and James Bay in rocky (Precambrian Shield) habitat, west to Lake Superior and Lake Athabasca. In Fig. 2, I placed the type locality symbol approximately in the middle of the range given by Richardson (1829). He said:

"Contrary to the practice of the Barren-Ground Caribou, the Woodland variety travels to the southward in the spring. They cross the Nelson and Severn Rivers [in Manitoba and Ontario, respectively] in immense herds in the month of May, pass the summer on the low, marshy shores of James’ Bay, and return to the northward, and … retire more inland in the month of September.”

Richardson (1829), although one of the greatest Arctic explorers of his era, had little experience with caribou south of the Barrenlands. He did not mention having examined museum specimens back in London, did not figure the animal or give metrics supporting its being "much larger than the Barren-Ground Caribou [and] has smaller horns” and most of his account of R. t. sylvestris is hearsay. His description of "immense herds” in migration precludes its application to woodland caribou.

Serrouya et al. (2012), using microsatellite markers, confirmed earlier conclusions (McLoughlin et al. 2004) that the caribou of the woodland ecotype DU6, sampled in Alberta north and south of the Peace River, clustered separately and were genetically distant from each other (average F_ST = 5.9% between pairs). Could those north of the Peace River be the newly discovered BEL clade of boreal "woodland” caribou in the Mackenzie Valley (cf. Polfus et al. 2017)?

The Atlantic-Gaspésie caribou (Fig. 2), DU11 (COSEWIC 2011), is an isolated montane ecotype (Pelletier et al. 2019). It is significantly differentiated from Labrador and boreal woodland caribou in Québec, F_ST 10.3%–17.2%, based on microsatellite loci (Courtois et al. 2003). Yannic et al. (2018), also based on microsatellite loci, gave its average F_ST from all other Canadian populations as 19%. This and its unique ecology (Pelletier et al. 2019; Frenette et al. 2020), warrant at least subspecific distinction. It remains unnamed.

Allen (1914) described the "Barren-ground Caribou of Labrador” from the Ungava Peninsula (Fig. 2) as Rangifer arcticus caboti Allen, 1914. Jacobi (1931) and Anderson (1946) thought it morphologically distinct enough for species status, R. caboti. Geist (1998) considered it a diagnostically "distinct form of barren-ground caribou” with antlers of "classical barren-ground form, but with short tines; brow and bez tines very close together; antlers usually widely spread”. Loughrey and Kelsall (1970) called it a migratory form of woodland caribou, R. c. caboti, and Cronin et al. (2005) confirmed that Labrador caribou share mtDNA haplotypes and have similar microsatellite allele frequencies to woodland caribou with ancient admixture from barren-ground caribou. There has been little recent introgression (Boulet et al. 2007; Klütsch et al. 2016; Taylor et al. 2021). It shares only one haplotype with barren-ground caribou and that is from the Baffin Island population (Cronin et al. 2005), itself unique as noted above.

Even though it is a currently accepted subspecies (Wilson and Mittermeier 2011), COSEWIC (2011) renamed it the Eastern Migratory ecotype, DU4 (Fig. 2). COSEWIC (2011) also extended its range to the west of Hudson’s Bay and James Bay, based on small migratory/aggregated calving populations at the south end of James Bay in Ontario and Quebec (Abraham and Thompson 1998; Brown et al. 2003) and west into Manitoba (Pond et al. 2016). Genetic evidence supports this extension (Klütsch et al. 2016).

COSEWIC (2011) further designated a montane ecotype in the Torngat Mountains as DU12, based on ecological differences from the migratory ecotype (Bergerud 2000; Boulet et al. 2007; Couturier et al. 2010). Using satellite tracking and microsatellite markers, Boulet et al. (2007) found no significant genetic differentiation (pairwise F_ST all ≤ 0.5%) between migratory and montane herds, whereas woodland caribou of southern Quebec were significantly differentiated (F_ST = 1.8%–4.8%). Yannic et al. (2016) confirmed the lack of genetic distinction between Torngat and migratory Labrador caribou. Labrador caribou are also smaller than Quebec woodland caribou (Couturier et al. 2010). Since the Torngat Mountain montane ecotype DU12 is not genetically distinct at the subspecies level from other Labrador caribou, it needs no formal description as a subspecies and Allen’s type specimen designation from the Torngat Mountains (Fig. 2) remains valid for Labrador caribou, R. c. caboti Allen, 1914.

By contrast, the "totally different” (Geist 2007) Newfoundland caribou, R. t. terraenovae Allen, 1896, currently a valid subspecies (Wilson and Mittermeier 2011; Wilkerson et al. 2018), has distinctive pelage and antlers of the "classical ‘woodland form’ but of large dimensions and spreading” (Geist 1991). Analysis of mtDNA haplotypes confirmed that they are NAL woodland caribou that likely diverged during the LGM in a coastal refugium (Cronin et al. 2005; Wilkerson et al. 2018; Yannic et al. 2018). The name R. caribou terraenovae Allen, 1896 (Jacobi 1931; Anderson 1946) is available and appropriate, given its genetic proximity and phylogenetic descent from other woodland caribou.