- Email: [email protected]
Polyphyletic Origin of the Small-Bodied, High-Arctic Subspecies of Tundra Reindeer (Rangifer tarandus)
MOLECULAR PHYLOGENETICS AND EVOLUTION
Vol. 10, No. 2, October, pp. 151–159, 1998 ARTICLE NO. FY980525
Polyphyletic Origin of the Small-Bodied, High-Arctic Subspecies of Tundra Reindeer (Rangifer tarandus) Peter Gravlund,*,1 Morten Meldgaard,† Svante Pa¨a¨bo,‡ and Peter Arctander* *Zoological Institute, Department of Population Biology, University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark; †Danish Polar Center, Strandgade 100H, DK-1401 Copenhagen K, Denmark; and ‡Zoological Institute, University of Munich, P.O. Box 202136, D-80021 Munich, Germany Received September 20, 1995; revised April 7, 1997
In order to investigate the origin of the three smallbodied, high-arctic subspecies of reindeer, Rangifer tarandus pearyi (the Canadian Archipelago), R. t. eogroenlandicus (East Greenland, extinct since 1900 AD), and R. t. platyrhynchus (Svalbard), samples were collected at nine localities from all six of the currently recognized subspecies of the tundra reindeer. A 203-bplong fragment of the mitochondrial control region was sequenced from 113 reindeer (Rangifer tarandus). The now extinct subspecies R. t. eogroenlandicus was for the first time included in a molecular study; DNA was extracted from four museum specimens (skins and bones) and successfully sequenced. A polyphyletic origin for the three subspecies of small-bodied, higharctic reindeer is suggested, with R. t. pearyi and R. t. eogroenlandicus being closely related and probably having evolved in high-arctic North America and R. t. platyrhynchus from Svalbard having evolved from Eurasian large-bodied reindeer. The small-bodied, higharctic reindeer presumably represent ecotypes that have evolved convergently in similar high-arctic environments. r 1998 Academic Press
INTRODUCTION The circumpolar distribution of the reindeer, Rangifer tarandus, spans a latitudinal range from 82°N at Ellesmere Island to 50°N in China (Banfield, 1961). The vast range of R. tarandus covers many different ecological biotopes, from high arctic islands through tundra to taiga woodlands (Leader-Williams, 1988). Based on both morphological characteristics of the different populations and biogeographical data, Banfield (1961) divided R. tarandus into two major groups:
1 To whom correspondance should be addressed. Fax: 35321010. E-mail: [email protected]. Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under Accession Nos. AF019570– AF19610.
tundra reindeer including six subspecies and the woodland reindeer including three subspecies. Emphasis has been placed on the need for studies to resolve the long-standing debate on whether the three small-bodied, high-arctic tundra reindeer subspecies, R. t. pearyi (from the Canadian Arctic Archipelago), R. t. platyrhynchus (from Svalbard), and the extinct R. t. eogroenlandicus (from East Greenland), have a monophyletic origin (Degerbøl, 1957; Banfield, 1961; Hakala et al., 1985; Røed, 1985; Røed et al., 1985; Røed et al., 1986). These subspecies all live (or lived) on high-arctic islands or other relatively isolated areas and are characterized by small body size, short legs, a short muzzle, and a paler, longer winter pelage relative to other R. tarandus subspecies (Degerbøl, 1957; Banfield, 1961). Similarities in distribution and morphology have led observers to propose two hypotheses to explain the evolution of these subspecies. The first suggests a monophyletic scenario in which small-bodied reindeer spread eastward from an origin in the Canadian Arctic Archipelago to Greenland and subsequently from Greenland to Svalbard (Hakala et al., 1985; Røed et al., 1986). The second hypothesis considers the small-bodied races of high-arctic reindeer to represent convergently evolved ecotypes adapted to similar high-arctic conditions. Specifically, this hypothesis suggests an ancestor–descendant relationship between R. t. pearyi and R. t. eogroenlandicus and the Svalbard population of R. t. platyrhychus as coming from the large-bodied R. t. tarandus of Novaya Zemlya (Russia), using Franz Joseph Land as a stepping stone. This hypothesis thus implies a diphyletic origin for the three subspecies (Hoel, 1916; Lønø, 1959; Willemsen, 1983). The ability to amplify and sequence DNA (Saiki et al., 1988; Gyllensten and Erlich, 1988) has made it possible to obtain sequences of specific DNA regions from a large number of individuals, and these methods are now increasingly used to study population variability. The capacity of the polymerase chain reaction to amplify minute quantities of DNA has furthermore enabled researchers to utilize ancient DNA sources, such as
151
1055-7903/98 $25.00 Copyright r 1998 by Academic Press All rights of reproduction in any form reserved.
152
GRAVLUND ET AL.
skins and bones from museum collections (see, e.g., Pa¨a¨bo et al., 1988; Hagelberg et al., 1989; Cooper et al., 1992). As a result, it is now possible to include in a molecular analysis of Rangifer subspecies samples from R. t. eogroenlandicus which became extinct circa 1900 (Degerbøl, 1957). In the present study, we address the following question: Do the three high-arctic, small-bodied subspecies have a monophyletic origin? To do so we compare sequences of a variable 203-bp-long region from the left domain of the control region of the mitochondrial genome (Saccone et al., 1991) to investigate intraspecific relationships among all six subspecies of tundra reindeer. This part of the control region has a high mutation rate, making it ideal for intraspecific comparisons (Brown, 1985). A phylogenetic maximum-parsimony analysis was carried out using elk (Alces alces) as outgroup. MATERIALS AND METHODS Sample Collection and Storage Included in this study were tissue and blood samples from 113 reindeer, comprising 109 specimens from extant populations and 4 specimens from an extinct population. The samples were obtained from nine localities around the circumpolar region, representing all six subspecies of the tundra reindeer group (sensu, Banfield, 1961). Distribution of the six subspecies and the nine sample localities are illustrated in Fig. 1, and location of collections, collectors, sample size, and subspecies are listed in Table 1. After collection, fresh tissue was preserved in saturated NaCl solution containing 25% dimethyl sulfoxide (Amos and Hoelzel, 1991) and subsequently stored at 280°C. Samples from specimens of the extinct subspecies R. t. eogroenlandicus (bones and skin samples) were processed in a room kept free of modern DNA and PCR products (a so-called ‘‘clean room’’) at the Zoological Institute, University of Munich. The four specimens of R. t. eogroenlandicus are stored in the collections of the Zoological Museum, University of Copenhagen, collection numbers: CN 647, CN 591, CN 592, and ZMK 9823. CN 591 and CN 592 are both dried skins from animals shot by the Ryder Expedition in 1892, Scoresbysund, East Greenland. Collection number CN 647 is a stray found antler of unknown age, with part of the skull attached. The DNA from this specimen was retrieved from pieces of dried skin on the skull. CN 647 was also collected by the Ryder Expedition. CN 9823 is an intact metatarsus containing dried marrow from an incomplete skeleton of unknown age found by the ‘‘Danmarks Expedition,’’ North-East Greenland, in 1907. As outgroup for the parsimony analysis we used Alces alces GenBank Accession No. U12866.
DNA Extraction DNA was obtained from tissue and blood samples by standard phenol/chloroform extraction. In order to obtain DNA from the skin and bone samples of the extinct specimens, the bone samples were ground and the dried skin samples were cut into minute, thin slices with a sterile scalpel. The DNA was extracted in a clean room, using a silica-based purification method (Ho¨ss and Pa¨a¨bo, 1993). PCR Amplification A 203-bp fragment from within the left domain of the mitochondrial control region was amplified by PCR (Saiki et al., 1988). Primers used were B16168H: 58-ggt tgc tgg ttt cac gcg gca tg-38 (Simonsen, pers. com., numbered according to the Bos taurus domesticus sequence, Anderson et al. (1982)) and RT1: 58-taa acg tac ata tat ggt cct gt-38. The latter primer was designed from reindeer sequences. The DNA from the modern samples was amplified in 50 µl reaction volumes containing 67 mM Tris–HCl, pH 8.8, 2 mM MgCl2, 16.6 mM (NH4 ) 2SO4, 10 mM b-mercaptoethanol, 0.2 mM dNTPs, 1 µM of each primer, and 0.4 units of Boehringer Mannheim Taq DNA polymerase. Amplification conditions were: 2 min of initial denaturation at 94°C for 1 cycle, followed by 1 min at 94°C, 1 min at 52°C, and 3 min at 72°C for 30 cycles. One microliter of the DNA-extraction was used as a template. The amplified double-stranded (ds) DNA was used as a template for further asymmetrical PCR reactions to produce single-stranded DNA for direct sequencing, using identical PCR conditions as for ds-amplification (Gyllensten and Erlich, 1988). For ancient samples, initial ds-DNA was amplified with a wax-mediated hotstart. The upper phase contained BSA (1.3 mg/ml), 0.75 units Taq Polymerase (Perkin–Elmer, Roche, NJ), as well as the DNA extract (5 µl). The lower phase contained 0.5 µM of each primer, 200 µM dNTPs, and 2 mM MgCl2. Amplification conditions were: 40 s denaturation at 92°C, followed by 40 cycles (1 min at 92°C, 1 min at 50°C, and 1 min at 72°C). To visualize the DNA, the products were separated by electrophoresis in a low-melting point agarose gel containing ethidium bromide. The amplified DNA was extracted from the agarose and diluted in 100 µl ddH2O, and 5 µl was used as template in subsequent 50 µl amplifications. Amplification products were then purified using Geneclean (BIO 101 Inc., La Jolla, CA) and used for direct sequencing as described by Bachmann et al. (1990). Direct Sequencing Single-stranded DNA from the extant specimens was concentrated by spin dialysis in Millipore Ultrafree-MC filters, type 30,000 NMWL. The sequencing reactions were performed with the dideoxy method (Sanger et al.,
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
153
FIG. 1. Map of the circumpolar region showing the distribution of the six subspecies of tundra reindeer and the sample localities: A 5 Rangifer tarandus tarandus [1 5 Norway, Riast Hyllingen (n 5 15); 2 5 the Taimyr Peninsula (n 5 23)]; B 5 R. t. granti Allen, 1902 [3 5 northeastern Alaska, the Porcupine Herd (n 5 7)]; C 5 R. t. groenlandicus [4 5 West Greenland, Kong Frederik d. IX. Land (n 5 14); 5 5 Baker Lake (n 5 8); 6 5 Baffin Island (n 5 7)]; D 5 R. t. pearyi [7a 5 Banks Island (n 5 5); 7b 5 Prince Patrick Island (n 5 2); 7c 5 Eglington Island (n 5 2); 7d 5 Melville Island (n 5 3); 7e 5 Prince of Wales Island (n 5 2); 7f 5 Somerset Island (n 5 1)], E 5 R. t. eogroenlandicus [8 5 East Greenland (n 5 4)], extinct since circa 1900; F 5 R. t. platyrhynchus [9 5 Svalbard, Nordenskiøld Land (n 5 20)]. Star marks the North Pole.
1977) using Sequenase kit ver. 2.0 (USB). Sequence products were separated by electrophoresis on 5% polyacrylamide gels. DNA sequences were verified by reading both strands (i.e., using both primers) and an overlap of approximately 150 nucleotides (75% overlap) was thus obtained. Population Subdivision and Genetic Diversity To determine whether the animals from the sampled locations belonged to different genetic populations a pairwise test for genetic differentiation between the compared samples was carried out, applying two different statistics, a haplotype-frequency-based (Hst) test and a sequence-based (Kst) test using the number of nucleotide differences among and between localities (Hudson et al., 1992). The significance of the observed Kst and Hst values was tested by means of Monte Carlo
simulations; individuals were randomly sampled from the compared localities, without replacement, and assigned to two new samples of the same size as the observed sets; the Kst and Hst values were then calculated. This was repeated 1000 times to obtain a distribution for the Kst and Hst values in the simulated sample sets. By comparing the observed values with the distribution of the simulated data sets, a probability of obtaining the observed values (or more extreme ones) was calculated (Hudson et al., 1992). The null hypothesis, i.e., that the observed and simulated data were genetically homogeneous, was rejected if the probability of the two data sets being similar was less than 0.05. Interpretation of these results was based upon arguments advanced by Hudson et al. (1992). If either of the two statistical methods
154
GRAVLUND ET AL.
rejected the null hypothesis, thus supporting heterogeneity between the compared localities, this conclusion was followed. If not, the two compared samples were treated as belonging to the same genetic population. The nucleotide diversity within and between the sample localities was estimated according to Nei (1987). This estimate is based on the average possibility of observing different bases at a nucleotide site in the comparisons of sequences. The interpopulational genetic divergence was corrected for intrapopulational diversity to obtain the net nucleotide substitution according to Nei (1987). Phylogenetic Analysis The reindeer mithochondrial haplotypes were aligned with the elk sequence by eye. All substitutions (transitions and transversions) were weighted equally. Gaps were treated so that each indel regardless of its length was equivalent to a single nucleotide substitution. The phylogenetic analyses were conducted using the program PAUP 3.1.1 (Swofford, 1993) carrying out a heuristic search using stepwise addition, performing tree-bisectionreconnection (TBR) branch swapping and employing the MULPARS function in effect. The elk sequence were obtained from GenBank Accession No. U12866. To test the branch support we applied the Support Index (Eernisse and Kluge, 1993). In the consensus tree of all the most parsimonious trees, the support of the persisting clades was found by collapsing all other nodes but the one in question. This was carried out in MacClade (Maddison and Maddison, 1992). The length of this new tree was determined by carrying out a constrained heuristic search in PAUP 3.1.1 (Swofford, 1993). The differences in length between the constrained tree and the most parsimonious trees indicate the number of characters supporting the clade NOT
TABLE 1
Sample size
Collecter/ origin
Norway
15
K. Røed
Taimyr Peninsula Alaska West Greenland Baker Lake Baffin Island Canadian Archipelago East Greenland
23 7 14 8 7 15
P. Gravlund R. White T. Romby Larsen C. Stroebeck C. Stroebeck C. Stroebeck
Svalbard
20
4
collapsed. This procedure is adapted from Eernisse and Kluge (1993) and was repeated for all clades found in the strict consensus tree of all the most parsimonious trees. RESULTS
Location, Size, Collector/Origin, and Subspecies of the Samples of Tundra Reindeer (Rangifer tarandus) Sample location
FIG. 2. The 41 haplotypes and the 36 variable sites defining them. ID numbers of the individuals are noted with their corresponding mtDNA haplotypes. Localities of the individuals are shown in bold: N 5 Norway, TY 5 Taimyr Peninsula, AK 5 Alaska, WG 5 West Greenland, BK 5 Baker Lake, BF 5 Baffin Island, CA 5 Canadian Archipelago, S 5 Svalbard. Specimens of the extinct East Greenland reindeer are noted with their collection numbers from the Zoological Museum, University of Copenhagen (CN and ZMK).
Subspecies R. t. tarandus (Domestic) R. t. tarandus. R. t. granti R. t. groenlandicus R. t. groenlandicus R. t. groenlandicus R. t. pearyi
ZMUC
R. t. eogroenlandicus K. Røed & I. Gjertz R. t. platyrhynchus
Note. ZMUC is Zoological Museum, University of Copenhagen and subspecies according to Banfield, 1961.
A 203-bp region of the mitochondrial control region was sequenced for 113 reindeer. Of the 203 bp, 36 sites (17.7%), comprising 33 transitions and 3 transversions, were found to be variable. Forty-one haplotypes could be recognized (Fig. 2), and of these, 23 were represented by only a single individual. No insertions or deletions were present between the reindeer haplotypes, and no sequence ambiguity was found. The outgroup elk (Alces alces) sequence was 204 bp long. When aligned with the reindeer it resulted in a 205-bp alignment. The alignment induced two indels; 1 bp in the elk sequence in position 5 and 2 bp in the reindeer sequences on positions 192–193. Test for Significant Genetic Heterogeneity The pairwise comparisons were conducted in a geographical consensus starting in Norway and going
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
155
eastward, comparing localities with adjacent localities (see Table 3). The first comparison between the two Eurasian localities, Norway and Taimyr, showed them to be genetically heterogeneous populations. Proceeding eastward, the pairwise comparisons Taimyr compared to Alaska and Alaska to Baker Lake both showed genetic heterogeneity. The next pairwise comparison, Baker Lake to Baffin Island, showed genetic homogeneity and these two localities were thereafter treated as one population (denoted as Baker Lake/Baffin Island). In all the following comparisons significant support for genetic heterogeneity was found: Baker Lake/Baffin Island with the Canadian Arctic Archipelago, the Canadian Arctic Archipelago with East Greenland, and Baker Lake/Baffin Island with West Greenland. The small-bodied, high-arctic reindeer on Svalbard were compared pairwise to the populations they were expected to possibly have been in contact with: East Greenland, Norway, and the Taimyr Peninsula. In all three comparisons significant support for genetic heterogeneity was found. Maximum Parsimony Analysis The heuristic search resulted in 1020 equally parsimonious trees, 94 steps long (consistency index 5 0.500 and retention index 5 0.682), based on 28 parsimoniously informative sites. A strict consensus tree (Fig. 3) constructed from these 1020 most parsimonious trees revealed seven basal clades (numbered 1–7 in Fig. 3) consisting of two to seven haplotypes. The support index for the seven clades is shown in Fig. 3. Clades 1 and 2 and 4–7 were found to be supported by one character. Clade 3 was found to be supported by three characters. In the cases where small- and large-bodied reindeer haplotypes come out as sister taxa, as in the case of haplotype 2 (small/large) and 3 (small) in clade 1, the support index is 2, and in the case of haplotype 32 (small) and 35 (large) in clade 6, the support index is 3. Genetic Diversity The 113 sampled reindeer had a nucleotide diversity of 3.42%. Intrapopulational nucleotide diversity varied considerably (Table 2), from a minimum of 0% in the West Greenland (n 5 14) and East Greenland samples (n 5 4) to a maximum of 3.36% and 3.42% in the Taimyr Peninsula (n 5 23) and Alaska samples (n 5 7), respectively. Net genetic diversity between the eight genetically distinct populations is presented in Table 2. The smallest net interpopulational diversity was found between the samples from the Canadian Arctic Archipelago and Baker Lake/Baffin Island (da 5 0.00092). The largest diversity was found between the samples from Svalbard and East Greenland and between the samples from Svalbard and West Greenland (in both cases da 5 0.02173). High genetic diversity was also found between the samples from Svalbard and the Canadian Arctic Archipelago (da 5 0.02013) and be-
FIG. 3. Strict consensus tree of the forty-one reindeer haplotypes (numbered as in Fig. 2) based on the 1020 most parsimonious trees found in heuristic search (L 5 1020; C.I. 5 0.500, including uninformative characters). The elk (Alces alces) was used as an outgroup. Seven basic clades (labeled 1 through 7) can be recognized. Numbers in parentheses at branches are the Support Index for that specific node. Asterisks indicates haplotypes found in the small-bodied, high-arctic reindeer.
tween West Greenland and Norway (da 5 0.02008). The samples from Svalbard had the smallest genetic diversity to Norway (da 5 0.01429) and to Taimyr (da 5 0.01637). DISCUSSION Origin of the Small-Bodied, High-Arctic Reindeer The parsimony analysis does not enable us to definitively answer the question of a polyphyletic origin of the small-bodied, high-arctic reindeer. As can be seen in Fig. 3, a firm resolution of all the taxa (haplotypes) is not possible. This is due to the relatively low number of informative sites (28) compared with the high number of taxa (41 reindeer haplotypes and 1 elk). But despite this, 7 clades (clades 1–7 in Fig. 3) can be recognized in all 1020 most parsimonious trees. In 2 of these clades (clades 1 and 6) we find haplotypes of both small-
156
GRAVLUND ET AL.
TABLE 2 Net Intra- and Interpopulational Diversity of Nine Populations of Reindeer (Rangifer tarandus) Based on a 203-Basepair-Fragment of the Mitochondrial Control Region
Norway (n 5 15) Taimyr (n 5 23) Alaska (n 5 7) West Greenland (n 5 14) Baker Lake (n 5 8) Baffin Island (n 5 7) Canadian Archipelago (n 5 15) East Greenland (n 5 4) Svalbard (n 5 20)
Norway
Taimyr
Alaska
0.0282
0.0055 0.0299
0.0119 0.0074 0.0338
West Greenland
Baker Lake
Baffin Island
Canadian Archipelago
East Greenland
0.0201 0.0111 0.0127 0.0000
0.0104 0.0027 0.0041 0.0037 0.0255
0.0122 0.0018 0.0069 0.0045 0.0000 0.0192
0.0154 0.0055 0.0081 0.0037 0.0010 0.0006 0.0114
0.0201 0.0137 0.0141 0.0099 0.0076 0.0101 0.0050 0.0000
Svalbard 0.0143 0.0164 0.0194 0.0217 0.0166 0.0196 0.0201 0.0217 0.0024
Note. Localities as in Fig. 1.
bodied, high-arctic and large reindeer. Both of these cases (support indices 2 and 3, respectively) demonstrate a substantial support for the existence of largeand small-bodied reindeer being each other’s nearest relatives and thereby suggests a polyphyletic origin of the small-bodied, high-arctic reindeer subspecies. If the small-bodied, high-arctic reindeer had a shared small-bodied ancestor we would have expected them to comprise a monophyletic clade in the cladogram. The mtDNA data thus point toward a di- or polyphyletic origin of the three small-bodied, high-arctic subspecies. A close correlation between R. t. pearyi and R. t. eogroenlandicus is suggested through the finding of a shared haplotype not found in the other populations (haplotype number 30 in Fig. 2). This haplotype is the only one found in the East Greenland samples and the most common haplotype in the samples from the Canadian Arctic Archipelago (C.A.A.), found in 6 out of 15 individuals. These mtDNA data correspond well with those of earlier investigations supporting a close relaTABLE 3 Pairwise Comparisons Testing for Genetic Heterogeneity Between Sample Locations Pairwise comparisons
Hst-value
Kst-value
Norway with Taimyr Taimyr with Alaska Alaska with Baker Lake Baker Lake with Baffin Island Baker L./Baffin Isl. With West Greenland Baker L./Baffin Isl. With C.A.A. C.A.A. with East Greenland Svalbard with East Greenland Svalbard with Norway Svalbard with Taimyr
0.057*** 0.014* 0.033ns 0.000ns 0.382*** 0.053*** 0.103ns 0.356*** 0.231*** 0.151***
0.084*** 0.082*** 0.069* 20.011ns 0.161*** 0.025ns 0.133* 0.762*** 0.349*** 0.327***
Note. Localities as found in Fig. 1. Hst is based on haplotype frequencies and Kst is based on nucleotide differences. Asterisks indicate level of significance: * 5 P , 0.05, ** 5 P , 0.01, *** 5 P , 0.001. ns means non significant.
tionship between R. t. pearyi and R. t. eogroenlandicus based on morphological as well as subfossil evidence and of an observation of animals migrating from Ellesmere Island to North Greenland in historical times (Roby et al., 1984). In answering the question of a monophyletic origin of all three subspecies of small-bodied, high-arctic subspecies, we are thus left with the key question of the origin of the Svalbard reindeer. Did they originate from small-bodied, high-arctic reindeer migrating to Svalbard from the west, Canada and/or East Greenland, or did they originate from large-bodied reindeer from Eurasia? The earliest indication of reindeer on Svalbard is the finding of reindeer droppings in peat cores dated to before 5000 yr BP (van der Knaap, 1986). Some workers have suggested a much longer history for reindeer in this area (Hakala et al., 1985), since parts of the islands are thought to have been unglaciated since 40,000 yr BP (Salvigsen, 1979). However, an ice-free area is not necessarily capable of sustaining a reindeer population. A reindeer population on Svalbard during the height of the last glaciation (20,000–18,000 yr BP) would have had to survive conditions comparable to those outlined in Meldgaard (1986) and in Meldgaard and Bennike (1989). They reject the possibility of a Weischlian refugium for reindeer in North Greenland/ Arctic Canada (Hakala et al., 1985), since the drier and colder conditions during that period would have turned the areas into arctic deserts, with insufficient forage biomass for a reindeer population to survive. However, as the climate became milder (within the last 10,000 years) reindeer could have become established on Svalbard. The parsimony analysis of the mtDNA sequence data supports an eastern origin for the Svalbard animals. In the strict consensus tree based on the 1020 most parsimonious trees (Fig. 3), the two haplotypes found in the Svalbard specimens (haplotypes 2 and 3 in Fig. 2)
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
and the haplotypes from the large-bodied reindeer from Norway (haplotypes 4 and 5 in Fig. 2) cluster together in clade 1. It should here be noted that haplotype 2 also is found in a reindeer from the Taimyr population. On the other hand it should also be noted that the support for clade 1 consists of only one character. However, neither of the Svalbard haplotypes were found in the samples from the two other small-bodied, high-arctic subspecies. An eastern origin of the Svalbard reindeer also corresponds well with the interpopulational net genetic diversity (Table 2), in which the Svalbard animals have the smallest difference from the samples from Norway and the Taimyr Peninsula (0.0143 and 0.0164, respectively) and the greatest difference from the individuals from East Greenland (R. t. eogroenlandicus) (0.0217), and West Greenland (R. t. groenlandicus) (0.0217) and to R. t. pearyi from the Canadian Arctic Archipelago (0.0201). Additional evidence points to an eastern origin for the Svalbard reindeer. Movement of reindeer from Novaya Zemlya to Svalbard is supported by the record of a reindeer shot on Svalbard in 1912 which had the foot of an ivory gull (Pagophila erburnea) tied to its left antler; the Samojed people of Novaja Zemlja are known to mark their reindeer by hanging objects from their antlers on certain occasions (Hoel, 1916). The archipelago of Franz Joseph Land has been proposed as a stepping stone on a migration from Russia to Svalbard (Lønø, 1959) and the finding of reindeer bones on Franz Joseph Land establishes that reindeer were present there some time during the Holocene (Bruce and Clarke, 1899). The size reduction that has apparently occurred in these subspecies is comparable to changes noted from fossil evidence in other large terrestrial mammals confined to islands. For example, male red deer (Cervus elaphus) on the island of Jersey decreased in size from approximately 200 to 36 kg in less than 6000 years (Lister, 1989) and mammoths (Mammuthus primigenius) on Wrangle Island in the East Siberian Sea, which survived well into the Holocene, showed a size reduction of 20–25% in less than 5000 years (Vartanyan et al., 1993). Explanations for such size reductions in large terrestrial mammals have focused on limited food resources during periods of high population density (Klein et al., 1987) and on climatic changes which alter the availability of food during critical growth periods (Guthrie, 1984). The former factor and the nutritional stress which accompanies it have been implicated in a reduction of up to 11% in shoulder height in 30–35 years in Russian reindeer (Klein et al., 1987). The latter effect has been elegantly demonstrated in experiments with Dall sheep (Ovis dalli) which show a positive correlation between body size, the quality and quantity of forage, and the length of the growth season (Guthrie, 1984). Under these artificial optimal conditions, the sheep grew considerably larger
157
than extant wild Dall Sheep, attaining the same size as Pleistocene specimens of this species (Guthrie, 1984). Thus, reduction in body size over a relatively short period is apparently not an uncommon phenomenon in mammalian species. Studies of transferrin variation have suggested a monophyletic origin of the small subspecies from Svalbard and the Canadian Archipelago (Røed et al., 1986). However, it should be noted that transferrin appears not to be independent of environmental factors because association has been reported between transferrin genotypes and environmental conditions, body weight (Røed et al., 1987), and susceptibility to epidemic and bacterial infections (Zhurkevich and Fomicheva, 1976, cited in Røed et al., 1987). The above is the interpretation of the present results that we find most probable. Alternative explanations could be a scenario where (i) the nucleotide differences on which the clades in Fig. 3 are based are homoplasies and therefore do not depict an evolutionary history or (ii) the small-bodied, high-arctic reindeer subspecies did indeed have a monophyletic origin and the observed pattern seen where small and large reindeer are sister taxa is the result of gene flow. The first alternative explanation cannot be rejected by the present results due to the limited support of the clades. A more firm resolution of the question can only be reached by additional sequencing in future studies. The second alternative explanation seems rather improbable considering the very recent history of reindeer on Svalbard and the fact that they share a haplotype with the Taimyr Peninsula reindeer even though the two groups are not believed to have been in contact in recent time. Nucleotide Diversity The nucleotide diversity in the 113 reindeer (3.42%) corresponds well to that found in the mitochondrial control region of other large land mammals. These studies have revealed nucleotide diversities of 4.10% for waterbuck (Kobus ellipsiprymnus, n 5 158, Simonsen, pers. com.) and 2.75% for impala (Aepyceros melampus, n 5 61, Arctander et al., 1996). Population History of Reindeer in Greenland Reflected in the mtDNA Data In both the West and East Greenland samples, the animals are found to have only one mitochondrial d-loop haplotype (the East Greenland samples were found to have only haplotype 30 and West Greenland only haplotype 1, see Fig. 2). The small genetic diversity may in part be the result of serious reductions of the population sizes (bottlenecks) (Nei, 1987), particularly since the effective population size of mitochondria is only one fourth of that of the nuclear genome due to the maternal inheritance and absence of recombination.
158
GRAVLUND ET AL.
The observed low level of genetic diversity in the samples from Greenland could be explained by: (i) a founder effect in which reindeer migrating to Greenland in the early Holocene would have had to cross the frozen Davis Strait from Baffin Island or cross the frozen Nares Strait from Ellesmere Island. Both migratory routes would have been difficult and it seems reasonable to assume that only relatively small groups of reindeer successfully completed the journey, possibly resulting in a strong founder effect and (ii) bottlenecks caused by the periodic population crashes. Severe population crashes of this kind are known from historical sources to have taken place in West Greenland at 65- to 125-year intervals (Meldgaard, 1986). One such bottleneck from Southwestern Greenland occurred during the 1970’s in which the total population, consisting of many small local ones, rapidly declined from approximately 100,000 individuals to approximately 8000 individuals (Roby et al., 1984). Continuously repeated bottlenecks would effectively reduce the genetic variation. CONCLUDING REMARKS The results of the mtDNA sequence analysis points to a diphyletic origin for small-bodied, high-arctic reindeer subspecies. This indicates that R. t. platyrhynchus and R. t. pearyi/eogroenlandicus are ecotypes of relatively recent origin, having evolved convergently, probably as a result of similar climatic and nutritional conditions. Very low mitochondrial genetic diversity is reported in the East and West Greenlandic reindeer populations and is thought to be caused by founder effect and bottlenecks. ACKNOWLEDGMENTS We thank the following people for making the samples available: Leonid Kolpatchikov, Knut Røed, Ian Gjerts and Bjørn Frantzen, Robert White, Curtis Stroebeck and Judith Eger, Troels Romby Larsen, and finally the Zoological Museum, University of Copenhagen. Ole Seberg, Bo Simonsen, and Per Kanneworff are thanked for computational assistance. A special thanks to Arne Redsted Rasmussen, Ole Seberg, Phill Clapham, Henrik Lund, Mary E. Petersen, and Per Palsbøl for constructive criticism and linguistic first aid. Economical support was provided by Axel Hemmingsens Legat, Clements Legat, Japetus Steenstrups Legat, and the Anne Nørremølle Foundation as well as the Carlsberg Foundation and the Danish Natural Science Research Counsil. Thanks to an anonymous reviewer for comments on the manuscript.
REFERENCES Arctander, P., Kat, P. W., Simonsen, B. T., and Siegismund, H. (1996). Population genetics of Kenyan impalas—consequences for conservation. In ‘‘Molecular Genetics in Conservation’’ (R. K. Wayne and T. B. Smith, Eds.), pp. 399–412. Oxford Univ. Press, Oxford. Amos, W., and Hoelzel, A. R. (1991). Long-term preservation of whale skin for DNA analysis. In ‘‘Genetic Ecology of Whales and Dolphins’’ (A. R. Hoelzel, Ed.), Reports of the International Whaling
Commission Special Issue 13, pp. 99–111. International Whaling Commission, Cambridge. Anderson, S., de Bruijn, M. H. L., Coulson, A. R., Eperson, I. C., Sanger, F., and Young, I. G. (1982). Complete sequence of bovine mitochondrial DNA. J. Mol. Biol. 156: 683–717. Bachman, B., Lu¨ke, W., and Hunsmann, G. (1990). Improvement of PCR amplified DNA sequencing with the aid of detergents. Nucleic Acids Res. 18: 1309. Banfield, A. W. F. (1961). A revision of the reindeer and caribou, genus Rangifer. National Museum Of Canada, Bulletin No. 177, Biological Series no. 66: 1–137. Brown, W. M. (1985). The mitochondrial genome of animals. In ‘‘Molecular Evolutionary Genetics’’ (MacIntyre, Ed.), pp. 95–130. Plenum, New York. Bruce, W. S., and Clarke, W. E. (1899). The Mammalia and birds of Franz Joseph Land. Proc. Roy. Phys. Soc. Edinbourg 16: 502–521. Cooper, A., Mourrer-Chauvire, C., Chambers, G. K., von Haeseler, A., Wilson, A. C., and Pa¨a¨bo, S. (1992). Independent origins of New Zealand moas and kiwis. Proc. Natl. Acad. Sci. USA 89: 8741– 8744. Degerbøl, M. (1957). The extinct reindeer of East Greenland. Acta Arctica 10: 57 pp. Eernisse, D. J., and Kluge, A. G. (1993). Taxonomic congruence versus total evidence, and amniote phylogeny inferred from fossils, molecules, and morphology. Mol. Biol. Evol. 10: 1170–1195. Guthrie, R. D. (1984). Alaskan megabucks, megabulls, and megarams: The issue of Pleistocene gigantism. Special Publication Carnegie Museum of Natural History. No. 8: 482–509. Gyllensten, U. B., and Erlich, H. A. (1988). Generation of singlestranded DNA by the polymerase chain reaction and its applications to direct sequencing of the HLA-DQA locus. Proc. Natl. Acad. Sci. USA 85: 7652–7656. Hagelberg, E., Sykes, B., and Hedges, R. (1989). Ancient bone DNA amplified. Nature 342: 485. Hakala, V. K., Staaland, H., Pulliainen, E., and Røed, K. H. (1985). Taxonomy and history of arctic island reindeer with special reference to Svalbard reindeer. Aquilo Ser. Zool. 23: 1–11. Hoel, A. (1916). Hvorfra er Spitsbergenrenen kommet?. Naturen 40: 37–43. [in Norwegian] Hudson, R. R., Boos, D. B., and Kapland, N. L. (1992). A statistical test for detecting geographical subdivision. Mol. Biol. Evol. 9: 138–151. Ho¨ss, M., and Pa¨a¨bo, S. (1993). DNA extraction from Pleistocene bones by a silica-based purification method. Proc. Natl. Acad. Sci. USA 21: 3913–3914. Klein, D. R., Meldgaard, M., and Fancy, S. G. (1987). Factors determinating leg length in Rangifer tarandus. J. Mammal. 68: 642–655. Knaap, van W. O. (1986). On the presence of reindeer Rangifer tarandus on Edgeøya, Spitsbergen in the period 3800–5000 B.P. Circumpolar J. 1: 3–10. Leader-Williams, N. (1988). Reindeer on South Georgia. Cambridge Univ. Press, Cambridge. Lister, A. M. (1989). Rapid dwarfing of red deer on Jersey in the last interglacial. Nature 342: 539–542. Lønø, O. (1959). Reinen på Svalbard. Naturen 12: 40–70. [In Norweigian] Maddison, W. P., and Maddison, D. R. (1992). ‘‘MacClade, Version 3.04.’’ Distributed by Sinauer Assosiates, Sunderland, MA. Meldgaard, M. (1986). The Greenland caribou—zoogeography, taxonomy and population dynamics. Meddr. Grønland. Biosci. 20: 1–88.
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER Meldgaard, M., and Bennike, O. (1989). Interglacial remains of caribou (Rangifer tarandus) and lemming (Dicrostonyx torquatus(?) ) from North Greenland. Boreas 18: 359–366. Nei, M. (1987). Molecular Evolutionary Genetics. Columbia Univ. Press, New York. Pa¨a¨bo, S., Gifford, J. A., and Wilson, A. C. (1988). Mitochondrial DNA sequences from a 7000-year old brain. Nucleic Acids Res. 16: 9775–9787. Roby, D. D., Thing, H., and Brink, K. L. (1984). History, status and taxanomic identity of caribou (Rangifer tarandus) in Northwest Greenland. Arctic 37: 23–30. Røed, K. (1985). Comparison of the genetic variation in Svalbard and Norwegian reindeer. Can. J. Zool. 63: 2038–2042. Røed, K., Soldal, A. V., and Thorisson, S. (1985). Transferrin variability and founder effect in Icelandic reindeer Rangifer tarandus L. Hereditas 103: 161–164. Røed, K., Staaland, H., Broughton, E., and Thomas, D. C. (1986). Transferrin variation in caribou (Rangifer tarandus L.) on the Canadian Arctic islands. Can. J. Zool. 64: 94–98. Røed, K. H., Mossing, T., Niemen, M., and Rydberg, A. (1987). Transferrin variation and genetic structure of reindeer populations in Scandinavia. Rangifer 7: 12–21.
159
Saccone, C., Pesole, G., and Sbisa´, E. (1991). The main regulatory region of mammalian mitochondrial DNA: Structure–function model and evolutionary pattern. J. Mol. Evol. 33: 83–91. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B., and Erlich, H. A. (1988). Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239: 487–491. Salvigsen, O. (1979). The last deglaciation of Svalbard. Boreas 8: 229–231. Sanger, F., Nichlen, S., and Coulson, A. R. (1977). DNA sequencing with chain-terminal inhibitors. Proc. Natl. Acad. Sci. USA 74: 5463–5468. Swofford, D. L. (1993). ‘‘PAUP Phylogenetic Analysis Using Parsimony Version 3.1,’’ Smithsonian Institution, Washington, DC. Vartanyan, S. L., Garutt, V. E., and Sher, A. V. (1993). Holocene dwarf mammoths from Wrangle Island in the Siberian Arctic. Nature 362: 337–340. Willemsen, G. F. (1983). Osteological measurements and some remarks on the evolution of the Svalbard reindeer Rangifer tarandus platyrhynchus. Z. Sa¨ugetierk. 48: 175–185. Zhurkevich, N. M., and Fomicheva, I. I. (1976). Genetic polymorphism of transferrin of blood serum in reindeer (Rangifer tarandus) indigenous to northeastern Siberia. (In Russian). Genetica 12: 56–65. (not seen).
Vol. 10, No. 2, October, pp. 151–159, 1998 ARTICLE NO. FY980525
Polyphyletic Origin of the Small-Bodied, High-Arctic Subspecies of Tundra Reindeer (Rangifer tarandus) Peter Gravlund,*,1 Morten Meldgaard,† Svante Pa¨a¨bo,‡ and Peter Arctander* *Zoological Institute, Department of Population Biology, University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark; †Danish Polar Center, Strandgade 100H, DK-1401 Copenhagen K, Denmark; and ‡Zoological Institute, University of Munich, P.O. Box 202136, D-80021 Munich, Germany Received September 20, 1995; revised April 7, 1997
In order to investigate the origin of the three smallbodied, high-arctic subspecies of reindeer, Rangifer tarandus pearyi (the Canadian Archipelago), R. t. eogroenlandicus (East Greenland, extinct since 1900 AD), and R. t. platyrhynchus (Svalbard), samples were collected at nine localities from all six of the currently recognized subspecies of the tundra reindeer. A 203-bplong fragment of the mitochondrial control region was sequenced from 113 reindeer (Rangifer tarandus). The now extinct subspecies R. t. eogroenlandicus was for the first time included in a molecular study; DNA was extracted from four museum specimens (skins and bones) and successfully sequenced. A polyphyletic origin for the three subspecies of small-bodied, higharctic reindeer is suggested, with R. t. pearyi and R. t. eogroenlandicus being closely related and probably having evolved in high-arctic North America and R. t. platyrhynchus from Svalbard having evolved from Eurasian large-bodied reindeer. The small-bodied, higharctic reindeer presumably represent ecotypes that have evolved convergently in similar high-arctic environments. r 1998 Academic Press
INTRODUCTION The circumpolar distribution of the reindeer, Rangifer tarandus, spans a latitudinal range from 82°N at Ellesmere Island to 50°N in China (Banfield, 1961). The vast range of R. tarandus covers many different ecological biotopes, from high arctic islands through tundra to taiga woodlands (Leader-Williams, 1988). Based on both morphological characteristics of the different populations and biogeographical data, Banfield (1961) divided R. tarandus into two major groups:
1 To whom correspondance should be addressed. Fax: 35321010. E-mail: [email protected]. Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under Accession Nos. AF019570– AF19610.
tundra reindeer including six subspecies and the woodland reindeer including three subspecies. Emphasis has been placed on the need for studies to resolve the long-standing debate on whether the three small-bodied, high-arctic tundra reindeer subspecies, R. t. pearyi (from the Canadian Arctic Archipelago), R. t. platyrhynchus (from Svalbard), and the extinct R. t. eogroenlandicus (from East Greenland), have a monophyletic origin (Degerbøl, 1957; Banfield, 1961; Hakala et al., 1985; Røed, 1985; Røed et al., 1985; Røed et al., 1986). These subspecies all live (or lived) on high-arctic islands or other relatively isolated areas and are characterized by small body size, short legs, a short muzzle, and a paler, longer winter pelage relative to other R. tarandus subspecies (Degerbøl, 1957; Banfield, 1961). Similarities in distribution and morphology have led observers to propose two hypotheses to explain the evolution of these subspecies. The first suggests a monophyletic scenario in which small-bodied reindeer spread eastward from an origin in the Canadian Arctic Archipelago to Greenland and subsequently from Greenland to Svalbard (Hakala et al., 1985; Røed et al., 1986). The second hypothesis considers the small-bodied races of high-arctic reindeer to represent convergently evolved ecotypes adapted to similar high-arctic conditions. Specifically, this hypothesis suggests an ancestor–descendant relationship between R. t. pearyi and R. t. eogroenlandicus and the Svalbard population of R. t. platyrhychus as coming from the large-bodied R. t. tarandus of Novaya Zemlya (Russia), using Franz Joseph Land as a stepping stone. This hypothesis thus implies a diphyletic origin for the three subspecies (Hoel, 1916; Lønø, 1959; Willemsen, 1983). The ability to amplify and sequence DNA (Saiki et al., 1988; Gyllensten and Erlich, 1988) has made it possible to obtain sequences of specific DNA regions from a large number of individuals, and these methods are now increasingly used to study population variability. The capacity of the polymerase chain reaction to amplify minute quantities of DNA has furthermore enabled researchers to utilize ancient DNA sources, such as
151
1055-7903/98 $25.00 Copyright r 1998 by Academic Press All rights of reproduction in any form reserved.
152
GRAVLUND ET AL.
skins and bones from museum collections (see, e.g., Pa¨a¨bo et al., 1988; Hagelberg et al., 1989; Cooper et al., 1992). As a result, it is now possible to include in a molecular analysis of Rangifer subspecies samples from R. t. eogroenlandicus which became extinct circa 1900 (Degerbøl, 1957). In the present study, we address the following question: Do the three high-arctic, small-bodied subspecies have a monophyletic origin? To do so we compare sequences of a variable 203-bp-long region from the left domain of the control region of the mitochondrial genome (Saccone et al., 1991) to investigate intraspecific relationships among all six subspecies of tundra reindeer. This part of the control region has a high mutation rate, making it ideal for intraspecific comparisons (Brown, 1985). A phylogenetic maximum-parsimony analysis was carried out using elk (Alces alces) as outgroup. MATERIALS AND METHODS Sample Collection and Storage Included in this study were tissue and blood samples from 113 reindeer, comprising 109 specimens from extant populations and 4 specimens from an extinct population. The samples were obtained from nine localities around the circumpolar region, representing all six subspecies of the tundra reindeer group (sensu, Banfield, 1961). Distribution of the six subspecies and the nine sample localities are illustrated in Fig. 1, and location of collections, collectors, sample size, and subspecies are listed in Table 1. After collection, fresh tissue was preserved in saturated NaCl solution containing 25% dimethyl sulfoxide (Amos and Hoelzel, 1991) and subsequently stored at 280°C. Samples from specimens of the extinct subspecies R. t. eogroenlandicus (bones and skin samples) were processed in a room kept free of modern DNA and PCR products (a so-called ‘‘clean room’’) at the Zoological Institute, University of Munich. The four specimens of R. t. eogroenlandicus are stored in the collections of the Zoological Museum, University of Copenhagen, collection numbers: CN 647, CN 591, CN 592, and ZMK 9823. CN 591 and CN 592 are both dried skins from animals shot by the Ryder Expedition in 1892, Scoresbysund, East Greenland. Collection number CN 647 is a stray found antler of unknown age, with part of the skull attached. The DNA from this specimen was retrieved from pieces of dried skin on the skull. CN 647 was also collected by the Ryder Expedition. CN 9823 is an intact metatarsus containing dried marrow from an incomplete skeleton of unknown age found by the ‘‘Danmarks Expedition,’’ North-East Greenland, in 1907. As outgroup for the parsimony analysis we used Alces alces GenBank Accession No. U12866.
DNA Extraction DNA was obtained from tissue and blood samples by standard phenol/chloroform extraction. In order to obtain DNA from the skin and bone samples of the extinct specimens, the bone samples were ground and the dried skin samples were cut into minute, thin slices with a sterile scalpel. The DNA was extracted in a clean room, using a silica-based purification method (Ho¨ss and Pa¨a¨bo, 1993). PCR Amplification A 203-bp fragment from within the left domain of the mitochondrial control region was amplified by PCR (Saiki et al., 1988). Primers used were B16168H: 58-ggt tgc tgg ttt cac gcg gca tg-38 (Simonsen, pers. com., numbered according to the Bos taurus domesticus sequence, Anderson et al. (1982)) and RT1: 58-taa acg tac ata tat ggt cct gt-38. The latter primer was designed from reindeer sequences. The DNA from the modern samples was amplified in 50 µl reaction volumes containing 67 mM Tris–HCl, pH 8.8, 2 mM MgCl2, 16.6 mM (NH4 ) 2SO4, 10 mM b-mercaptoethanol, 0.2 mM dNTPs, 1 µM of each primer, and 0.4 units of Boehringer Mannheim Taq DNA polymerase. Amplification conditions were: 2 min of initial denaturation at 94°C for 1 cycle, followed by 1 min at 94°C, 1 min at 52°C, and 3 min at 72°C for 30 cycles. One microliter of the DNA-extraction was used as a template. The amplified double-stranded (ds) DNA was used as a template for further asymmetrical PCR reactions to produce single-stranded DNA for direct sequencing, using identical PCR conditions as for ds-amplification (Gyllensten and Erlich, 1988). For ancient samples, initial ds-DNA was amplified with a wax-mediated hotstart. The upper phase contained BSA (1.3 mg/ml), 0.75 units Taq Polymerase (Perkin–Elmer, Roche, NJ), as well as the DNA extract (5 µl). The lower phase contained 0.5 µM of each primer, 200 µM dNTPs, and 2 mM MgCl2. Amplification conditions were: 40 s denaturation at 92°C, followed by 40 cycles (1 min at 92°C, 1 min at 50°C, and 1 min at 72°C). To visualize the DNA, the products were separated by electrophoresis in a low-melting point agarose gel containing ethidium bromide. The amplified DNA was extracted from the agarose and diluted in 100 µl ddH2O, and 5 µl was used as template in subsequent 50 µl amplifications. Amplification products were then purified using Geneclean (BIO 101 Inc., La Jolla, CA) and used for direct sequencing as described by Bachmann et al. (1990). Direct Sequencing Single-stranded DNA from the extant specimens was concentrated by spin dialysis in Millipore Ultrafree-MC filters, type 30,000 NMWL. The sequencing reactions were performed with the dideoxy method (Sanger et al.,
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
153
FIG. 1. Map of the circumpolar region showing the distribution of the six subspecies of tundra reindeer and the sample localities: A 5 Rangifer tarandus tarandus [1 5 Norway, Riast Hyllingen (n 5 15); 2 5 the Taimyr Peninsula (n 5 23)]; B 5 R. t. granti Allen, 1902 [3 5 northeastern Alaska, the Porcupine Herd (n 5 7)]; C 5 R. t. groenlandicus [4 5 West Greenland, Kong Frederik d. IX. Land (n 5 14); 5 5 Baker Lake (n 5 8); 6 5 Baffin Island (n 5 7)]; D 5 R. t. pearyi [7a 5 Banks Island (n 5 5); 7b 5 Prince Patrick Island (n 5 2); 7c 5 Eglington Island (n 5 2); 7d 5 Melville Island (n 5 3); 7e 5 Prince of Wales Island (n 5 2); 7f 5 Somerset Island (n 5 1)], E 5 R. t. eogroenlandicus [8 5 East Greenland (n 5 4)], extinct since circa 1900; F 5 R. t. platyrhynchus [9 5 Svalbard, Nordenskiøld Land (n 5 20)]. Star marks the North Pole.
1977) using Sequenase kit ver. 2.0 (USB). Sequence products were separated by electrophoresis on 5% polyacrylamide gels. DNA sequences were verified by reading both strands (i.e., using both primers) and an overlap of approximately 150 nucleotides (75% overlap) was thus obtained. Population Subdivision and Genetic Diversity To determine whether the animals from the sampled locations belonged to different genetic populations a pairwise test for genetic differentiation between the compared samples was carried out, applying two different statistics, a haplotype-frequency-based (Hst) test and a sequence-based (Kst) test using the number of nucleotide differences among and between localities (Hudson et al., 1992). The significance of the observed Kst and Hst values was tested by means of Monte Carlo
simulations; individuals were randomly sampled from the compared localities, without replacement, and assigned to two new samples of the same size as the observed sets; the Kst and Hst values were then calculated. This was repeated 1000 times to obtain a distribution for the Kst and Hst values in the simulated sample sets. By comparing the observed values with the distribution of the simulated data sets, a probability of obtaining the observed values (or more extreme ones) was calculated (Hudson et al., 1992). The null hypothesis, i.e., that the observed and simulated data were genetically homogeneous, was rejected if the probability of the two data sets being similar was less than 0.05. Interpretation of these results was based upon arguments advanced by Hudson et al. (1992). If either of the two statistical methods
154
GRAVLUND ET AL.
rejected the null hypothesis, thus supporting heterogeneity between the compared localities, this conclusion was followed. If not, the two compared samples were treated as belonging to the same genetic population. The nucleotide diversity within and between the sample localities was estimated according to Nei (1987). This estimate is based on the average possibility of observing different bases at a nucleotide site in the comparisons of sequences. The interpopulational genetic divergence was corrected for intrapopulational diversity to obtain the net nucleotide substitution according to Nei (1987). Phylogenetic Analysis The reindeer mithochondrial haplotypes were aligned with the elk sequence by eye. All substitutions (transitions and transversions) were weighted equally. Gaps were treated so that each indel regardless of its length was equivalent to a single nucleotide substitution. The phylogenetic analyses were conducted using the program PAUP 3.1.1 (Swofford, 1993) carrying out a heuristic search using stepwise addition, performing tree-bisectionreconnection (TBR) branch swapping and employing the MULPARS function in effect. The elk sequence were obtained from GenBank Accession No. U12866. To test the branch support we applied the Support Index (Eernisse and Kluge, 1993). In the consensus tree of all the most parsimonious trees, the support of the persisting clades was found by collapsing all other nodes but the one in question. This was carried out in MacClade (Maddison and Maddison, 1992). The length of this new tree was determined by carrying out a constrained heuristic search in PAUP 3.1.1 (Swofford, 1993). The differences in length between the constrained tree and the most parsimonious trees indicate the number of characters supporting the clade NOT
TABLE 1
Sample size
Collecter/ origin
Norway
15
K. Røed
Taimyr Peninsula Alaska West Greenland Baker Lake Baffin Island Canadian Archipelago East Greenland
23 7 14 8 7 15
P. Gravlund R. White T. Romby Larsen C. Stroebeck C. Stroebeck C. Stroebeck
Svalbard
20
4
collapsed. This procedure is adapted from Eernisse and Kluge (1993) and was repeated for all clades found in the strict consensus tree of all the most parsimonious trees. RESULTS
Location, Size, Collector/Origin, and Subspecies of the Samples of Tundra Reindeer (Rangifer tarandus) Sample location
FIG. 2. The 41 haplotypes and the 36 variable sites defining them. ID numbers of the individuals are noted with their corresponding mtDNA haplotypes. Localities of the individuals are shown in bold: N 5 Norway, TY 5 Taimyr Peninsula, AK 5 Alaska, WG 5 West Greenland, BK 5 Baker Lake, BF 5 Baffin Island, CA 5 Canadian Archipelago, S 5 Svalbard. Specimens of the extinct East Greenland reindeer are noted with their collection numbers from the Zoological Museum, University of Copenhagen (CN and ZMK).
Subspecies R. t. tarandus (Domestic) R. t. tarandus. R. t. granti R. t. groenlandicus R. t. groenlandicus R. t. groenlandicus R. t. pearyi
ZMUC
R. t. eogroenlandicus K. Røed & I. Gjertz R. t. platyrhynchus
Note. ZMUC is Zoological Museum, University of Copenhagen and subspecies according to Banfield, 1961.
A 203-bp region of the mitochondrial control region was sequenced for 113 reindeer. Of the 203 bp, 36 sites (17.7%), comprising 33 transitions and 3 transversions, were found to be variable. Forty-one haplotypes could be recognized (Fig. 2), and of these, 23 were represented by only a single individual. No insertions or deletions were present between the reindeer haplotypes, and no sequence ambiguity was found. The outgroup elk (Alces alces) sequence was 204 bp long. When aligned with the reindeer it resulted in a 205-bp alignment. The alignment induced two indels; 1 bp in the elk sequence in position 5 and 2 bp in the reindeer sequences on positions 192–193. Test for Significant Genetic Heterogeneity The pairwise comparisons were conducted in a geographical consensus starting in Norway and going
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
155
eastward, comparing localities with adjacent localities (see Table 3). The first comparison between the two Eurasian localities, Norway and Taimyr, showed them to be genetically heterogeneous populations. Proceeding eastward, the pairwise comparisons Taimyr compared to Alaska and Alaska to Baker Lake both showed genetic heterogeneity. The next pairwise comparison, Baker Lake to Baffin Island, showed genetic homogeneity and these two localities were thereafter treated as one population (denoted as Baker Lake/Baffin Island). In all the following comparisons significant support for genetic heterogeneity was found: Baker Lake/Baffin Island with the Canadian Arctic Archipelago, the Canadian Arctic Archipelago with East Greenland, and Baker Lake/Baffin Island with West Greenland. The small-bodied, high-arctic reindeer on Svalbard were compared pairwise to the populations they were expected to possibly have been in contact with: East Greenland, Norway, and the Taimyr Peninsula. In all three comparisons significant support for genetic heterogeneity was found. Maximum Parsimony Analysis The heuristic search resulted in 1020 equally parsimonious trees, 94 steps long (consistency index 5 0.500 and retention index 5 0.682), based on 28 parsimoniously informative sites. A strict consensus tree (Fig. 3) constructed from these 1020 most parsimonious trees revealed seven basal clades (numbered 1–7 in Fig. 3) consisting of two to seven haplotypes. The support index for the seven clades is shown in Fig. 3. Clades 1 and 2 and 4–7 were found to be supported by one character. Clade 3 was found to be supported by three characters. In the cases where small- and large-bodied reindeer haplotypes come out as sister taxa, as in the case of haplotype 2 (small/large) and 3 (small) in clade 1, the support index is 2, and in the case of haplotype 32 (small) and 35 (large) in clade 6, the support index is 3. Genetic Diversity The 113 sampled reindeer had a nucleotide diversity of 3.42%. Intrapopulational nucleotide diversity varied considerably (Table 2), from a minimum of 0% in the West Greenland (n 5 14) and East Greenland samples (n 5 4) to a maximum of 3.36% and 3.42% in the Taimyr Peninsula (n 5 23) and Alaska samples (n 5 7), respectively. Net genetic diversity between the eight genetically distinct populations is presented in Table 2. The smallest net interpopulational diversity was found between the samples from the Canadian Arctic Archipelago and Baker Lake/Baffin Island (da 5 0.00092). The largest diversity was found between the samples from Svalbard and East Greenland and between the samples from Svalbard and West Greenland (in both cases da 5 0.02173). High genetic diversity was also found between the samples from Svalbard and the Canadian Arctic Archipelago (da 5 0.02013) and be-
FIG. 3. Strict consensus tree of the forty-one reindeer haplotypes (numbered as in Fig. 2) based on the 1020 most parsimonious trees found in heuristic search (L 5 1020; C.I. 5 0.500, including uninformative characters). The elk (Alces alces) was used as an outgroup. Seven basic clades (labeled 1 through 7) can be recognized. Numbers in parentheses at branches are the Support Index for that specific node. Asterisks indicates haplotypes found in the small-bodied, high-arctic reindeer.
tween West Greenland and Norway (da 5 0.02008). The samples from Svalbard had the smallest genetic diversity to Norway (da 5 0.01429) and to Taimyr (da 5 0.01637). DISCUSSION Origin of the Small-Bodied, High-Arctic Reindeer The parsimony analysis does not enable us to definitively answer the question of a polyphyletic origin of the small-bodied, high-arctic reindeer. As can be seen in Fig. 3, a firm resolution of all the taxa (haplotypes) is not possible. This is due to the relatively low number of informative sites (28) compared with the high number of taxa (41 reindeer haplotypes and 1 elk). But despite this, 7 clades (clades 1–7 in Fig. 3) can be recognized in all 1020 most parsimonious trees. In 2 of these clades (clades 1 and 6) we find haplotypes of both small-
156
GRAVLUND ET AL.
TABLE 2 Net Intra- and Interpopulational Diversity of Nine Populations of Reindeer (Rangifer tarandus) Based on a 203-Basepair-Fragment of the Mitochondrial Control Region
Norway (n 5 15) Taimyr (n 5 23) Alaska (n 5 7) West Greenland (n 5 14) Baker Lake (n 5 8) Baffin Island (n 5 7) Canadian Archipelago (n 5 15) East Greenland (n 5 4) Svalbard (n 5 20)
Norway
Taimyr
Alaska
0.0282
0.0055 0.0299
0.0119 0.0074 0.0338
West Greenland
Baker Lake
Baffin Island
Canadian Archipelago
East Greenland
0.0201 0.0111 0.0127 0.0000
0.0104 0.0027 0.0041 0.0037 0.0255
0.0122 0.0018 0.0069 0.0045 0.0000 0.0192
0.0154 0.0055 0.0081 0.0037 0.0010 0.0006 0.0114
0.0201 0.0137 0.0141 0.0099 0.0076 0.0101 0.0050 0.0000
Svalbard 0.0143 0.0164 0.0194 0.0217 0.0166 0.0196 0.0201 0.0217 0.0024
Note. Localities as in Fig. 1.
bodied, high-arctic and large reindeer. Both of these cases (support indices 2 and 3, respectively) demonstrate a substantial support for the existence of largeand small-bodied reindeer being each other’s nearest relatives and thereby suggests a polyphyletic origin of the small-bodied, high-arctic reindeer subspecies. If the small-bodied, high-arctic reindeer had a shared small-bodied ancestor we would have expected them to comprise a monophyletic clade in the cladogram. The mtDNA data thus point toward a di- or polyphyletic origin of the three small-bodied, high-arctic subspecies. A close correlation between R. t. pearyi and R. t. eogroenlandicus is suggested through the finding of a shared haplotype not found in the other populations (haplotype number 30 in Fig. 2). This haplotype is the only one found in the East Greenland samples and the most common haplotype in the samples from the Canadian Arctic Archipelago (C.A.A.), found in 6 out of 15 individuals. These mtDNA data correspond well with those of earlier investigations supporting a close relaTABLE 3 Pairwise Comparisons Testing for Genetic Heterogeneity Between Sample Locations Pairwise comparisons
Hst-value
Kst-value
Norway with Taimyr Taimyr with Alaska Alaska with Baker Lake Baker Lake with Baffin Island Baker L./Baffin Isl. With West Greenland Baker L./Baffin Isl. With C.A.A. C.A.A. with East Greenland Svalbard with East Greenland Svalbard with Norway Svalbard with Taimyr
0.057*** 0.014* 0.033ns 0.000ns 0.382*** 0.053*** 0.103ns 0.356*** 0.231*** 0.151***
0.084*** 0.082*** 0.069* 20.011ns 0.161*** 0.025ns 0.133* 0.762*** 0.349*** 0.327***
Note. Localities as found in Fig. 1. Hst is based on haplotype frequencies and Kst is based on nucleotide differences. Asterisks indicate level of significance: * 5 P , 0.05, ** 5 P , 0.01, *** 5 P , 0.001. ns means non significant.
tionship between R. t. pearyi and R. t. eogroenlandicus based on morphological as well as subfossil evidence and of an observation of animals migrating from Ellesmere Island to North Greenland in historical times (Roby et al., 1984). In answering the question of a monophyletic origin of all three subspecies of small-bodied, high-arctic subspecies, we are thus left with the key question of the origin of the Svalbard reindeer. Did they originate from small-bodied, high-arctic reindeer migrating to Svalbard from the west, Canada and/or East Greenland, or did they originate from large-bodied reindeer from Eurasia? The earliest indication of reindeer on Svalbard is the finding of reindeer droppings in peat cores dated to before 5000 yr BP (van der Knaap, 1986). Some workers have suggested a much longer history for reindeer in this area (Hakala et al., 1985), since parts of the islands are thought to have been unglaciated since 40,000 yr BP (Salvigsen, 1979). However, an ice-free area is not necessarily capable of sustaining a reindeer population. A reindeer population on Svalbard during the height of the last glaciation (20,000–18,000 yr BP) would have had to survive conditions comparable to those outlined in Meldgaard (1986) and in Meldgaard and Bennike (1989). They reject the possibility of a Weischlian refugium for reindeer in North Greenland/ Arctic Canada (Hakala et al., 1985), since the drier and colder conditions during that period would have turned the areas into arctic deserts, with insufficient forage biomass for a reindeer population to survive. However, as the climate became milder (within the last 10,000 years) reindeer could have become established on Svalbard. The parsimony analysis of the mtDNA sequence data supports an eastern origin for the Svalbard animals. In the strict consensus tree based on the 1020 most parsimonious trees (Fig. 3), the two haplotypes found in the Svalbard specimens (haplotypes 2 and 3 in Fig. 2)
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER
and the haplotypes from the large-bodied reindeer from Norway (haplotypes 4 and 5 in Fig. 2) cluster together in clade 1. It should here be noted that haplotype 2 also is found in a reindeer from the Taimyr population. On the other hand it should also be noted that the support for clade 1 consists of only one character. However, neither of the Svalbard haplotypes were found in the samples from the two other small-bodied, high-arctic subspecies. An eastern origin of the Svalbard reindeer also corresponds well with the interpopulational net genetic diversity (Table 2), in which the Svalbard animals have the smallest difference from the samples from Norway and the Taimyr Peninsula (0.0143 and 0.0164, respectively) and the greatest difference from the individuals from East Greenland (R. t. eogroenlandicus) (0.0217), and West Greenland (R. t. groenlandicus) (0.0217) and to R. t. pearyi from the Canadian Arctic Archipelago (0.0201). Additional evidence points to an eastern origin for the Svalbard reindeer. Movement of reindeer from Novaya Zemlya to Svalbard is supported by the record of a reindeer shot on Svalbard in 1912 which had the foot of an ivory gull (Pagophila erburnea) tied to its left antler; the Samojed people of Novaja Zemlja are known to mark their reindeer by hanging objects from their antlers on certain occasions (Hoel, 1916). The archipelago of Franz Joseph Land has been proposed as a stepping stone on a migration from Russia to Svalbard (Lønø, 1959) and the finding of reindeer bones on Franz Joseph Land establishes that reindeer were present there some time during the Holocene (Bruce and Clarke, 1899). The size reduction that has apparently occurred in these subspecies is comparable to changes noted from fossil evidence in other large terrestrial mammals confined to islands. For example, male red deer (Cervus elaphus) on the island of Jersey decreased in size from approximately 200 to 36 kg in less than 6000 years (Lister, 1989) and mammoths (Mammuthus primigenius) on Wrangle Island in the East Siberian Sea, which survived well into the Holocene, showed a size reduction of 20–25% in less than 5000 years (Vartanyan et al., 1993). Explanations for such size reductions in large terrestrial mammals have focused on limited food resources during periods of high population density (Klein et al., 1987) and on climatic changes which alter the availability of food during critical growth periods (Guthrie, 1984). The former factor and the nutritional stress which accompanies it have been implicated in a reduction of up to 11% in shoulder height in 30–35 years in Russian reindeer (Klein et al., 1987). The latter effect has been elegantly demonstrated in experiments with Dall sheep (Ovis dalli) which show a positive correlation between body size, the quality and quantity of forage, and the length of the growth season (Guthrie, 1984). Under these artificial optimal conditions, the sheep grew considerably larger
157
than extant wild Dall Sheep, attaining the same size as Pleistocene specimens of this species (Guthrie, 1984). Thus, reduction in body size over a relatively short period is apparently not an uncommon phenomenon in mammalian species. Studies of transferrin variation have suggested a monophyletic origin of the small subspecies from Svalbard and the Canadian Archipelago (Røed et al., 1986). However, it should be noted that transferrin appears not to be independent of environmental factors because association has been reported between transferrin genotypes and environmental conditions, body weight (Røed et al., 1987), and susceptibility to epidemic and bacterial infections (Zhurkevich and Fomicheva, 1976, cited in Røed et al., 1987). The above is the interpretation of the present results that we find most probable. Alternative explanations could be a scenario where (i) the nucleotide differences on which the clades in Fig. 3 are based are homoplasies and therefore do not depict an evolutionary history or (ii) the small-bodied, high-arctic reindeer subspecies did indeed have a monophyletic origin and the observed pattern seen where small and large reindeer are sister taxa is the result of gene flow. The first alternative explanation cannot be rejected by the present results due to the limited support of the clades. A more firm resolution of the question can only be reached by additional sequencing in future studies. The second alternative explanation seems rather improbable considering the very recent history of reindeer on Svalbard and the fact that they share a haplotype with the Taimyr Peninsula reindeer even though the two groups are not believed to have been in contact in recent time. Nucleotide Diversity The nucleotide diversity in the 113 reindeer (3.42%) corresponds well to that found in the mitochondrial control region of other large land mammals. These studies have revealed nucleotide diversities of 4.10% for waterbuck (Kobus ellipsiprymnus, n 5 158, Simonsen, pers. com.) and 2.75% for impala (Aepyceros melampus, n 5 61, Arctander et al., 1996). Population History of Reindeer in Greenland Reflected in the mtDNA Data In both the West and East Greenland samples, the animals are found to have only one mitochondrial d-loop haplotype (the East Greenland samples were found to have only haplotype 30 and West Greenland only haplotype 1, see Fig. 2). The small genetic diversity may in part be the result of serious reductions of the population sizes (bottlenecks) (Nei, 1987), particularly since the effective population size of mitochondria is only one fourth of that of the nuclear genome due to the maternal inheritance and absence of recombination.
158
GRAVLUND ET AL.
The observed low level of genetic diversity in the samples from Greenland could be explained by: (i) a founder effect in which reindeer migrating to Greenland in the early Holocene would have had to cross the frozen Davis Strait from Baffin Island or cross the frozen Nares Strait from Ellesmere Island. Both migratory routes would have been difficult and it seems reasonable to assume that only relatively small groups of reindeer successfully completed the journey, possibly resulting in a strong founder effect and (ii) bottlenecks caused by the periodic population crashes. Severe population crashes of this kind are known from historical sources to have taken place in West Greenland at 65- to 125-year intervals (Meldgaard, 1986). One such bottleneck from Southwestern Greenland occurred during the 1970’s in which the total population, consisting of many small local ones, rapidly declined from approximately 100,000 individuals to approximately 8000 individuals (Roby et al., 1984). Continuously repeated bottlenecks would effectively reduce the genetic variation. CONCLUDING REMARKS The results of the mtDNA sequence analysis points to a diphyletic origin for small-bodied, high-arctic reindeer subspecies. This indicates that R. t. platyrhynchus and R. t. pearyi/eogroenlandicus are ecotypes of relatively recent origin, having evolved convergently, probably as a result of similar climatic and nutritional conditions. Very low mitochondrial genetic diversity is reported in the East and West Greenlandic reindeer populations and is thought to be caused by founder effect and bottlenecks. ACKNOWLEDGMENTS We thank the following people for making the samples available: Leonid Kolpatchikov, Knut Røed, Ian Gjerts and Bjørn Frantzen, Robert White, Curtis Stroebeck and Judith Eger, Troels Romby Larsen, and finally the Zoological Museum, University of Copenhagen. Ole Seberg, Bo Simonsen, and Per Kanneworff are thanked for computational assistance. A special thanks to Arne Redsted Rasmussen, Ole Seberg, Phill Clapham, Henrik Lund, Mary E. Petersen, and Per Palsbøl for constructive criticism and linguistic first aid. Economical support was provided by Axel Hemmingsens Legat, Clements Legat, Japetus Steenstrups Legat, and the Anne Nørremølle Foundation as well as the Carlsberg Foundation and the Danish Natural Science Research Counsil. Thanks to an anonymous reviewer for comments on the manuscript.
REFERENCES Arctander, P., Kat, P. W., Simonsen, B. T., and Siegismund, H. (1996). Population genetics of Kenyan impalas—consequences for conservation. In ‘‘Molecular Genetics in Conservation’’ (R. K. Wayne and T. B. Smith, Eds.), pp. 399–412. Oxford Univ. Press, Oxford. Amos, W., and Hoelzel, A. R. (1991). Long-term preservation of whale skin for DNA analysis. In ‘‘Genetic Ecology of Whales and Dolphins’’ (A. R. Hoelzel, Ed.), Reports of the International Whaling
Commission Special Issue 13, pp. 99–111. International Whaling Commission, Cambridge. Anderson, S., de Bruijn, M. H. L., Coulson, A. R., Eperson, I. C., Sanger, F., and Young, I. G. (1982). Complete sequence of bovine mitochondrial DNA. J. Mol. Biol. 156: 683–717. Bachman, B., Lu¨ke, W., and Hunsmann, G. (1990). Improvement of PCR amplified DNA sequencing with the aid of detergents. Nucleic Acids Res. 18: 1309. Banfield, A. W. F. (1961). A revision of the reindeer and caribou, genus Rangifer. National Museum Of Canada, Bulletin No. 177, Biological Series no. 66: 1–137. Brown, W. M. (1985). The mitochondrial genome of animals. In ‘‘Molecular Evolutionary Genetics’’ (MacIntyre, Ed.), pp. 95–130. Plenum, New York. Bruce, W. S., and Clarke, W. E. (1899). The Mammalia and birds of Franz Joseph Land. Proc. Roy. Phys. Soc. Edinbourg 16: 502–521. Cooper, A., Mourrer-Chauvire, C., Chambers, G. K., von Haeseler, A., Wilson, A. C., and Pa¨a¨bo, S. (1992). Independent origins of New Zealand moas and kiwis. Proc. Natl. Acad. Sci. USA 89: 8741– 8744. Degerbøl, M. (1957). The extinct reindeer of East Greenland. Acta Arctica 10: 57 pp. Eernisse, D. J., and Kluge, A. G. (1993). Taxonomic congruence versus total evidence, and amniote phylogeny inferred from fossils, molecules, and morphology. Mol. Biol. Evol. 10: 1170–1195. Guthrie, R. D. (1984). Alaskan megabucks, megabulls, and megarams: The issue of Pleistocene gigantism. Special Publication Carnegie Museum of Natural History. No. 8: 482–509. Gyllensten, U. B., and Erlich, H. A. (1988). Generation of singlestranded DNA by the polymerase chain reaction and its applications to direct sequencing of the HLA-DQA locus. Proc. Natl. Acad. Sci. USA 85: 7652–7656. Hagelberg, E., Sykes, B., and Hedges, R. (1989). Ancient bone DNA amplified. Nature 342: 485. Hakala, V. K., Staaland, H., Pulliainen, E., and Røed, K. H. (1985). Taxonomy and history of arctic island reindeer with special reference to Svalbard reindeer. Aquilo Ser. Zool. 23: 1–11. Hoel, A. (1916). Hvorfra er Spitsbergenrenen kommet?. Naturen 40: 37–43. [in Norwegian] Hudson, R. R., Boos, D. B., and Kapland, N. L. (1992). A statistical test for detecting geographical subdivision. Mol. Biol. Evol. 9: 138–151. Ho¨ss, M., and Pa¨a¨bo, S. (1993). DNA extraction from Pleistocene bones by a silica-based purification method. Proc. Natl. Acad. Sci. USA 21: 3913–3914. Klein, D. R., Meldgaard, M., and Fancy, S. G. (1987). Factors determinating leg length in Rangifer tarandus. J. Mammal. 68: 642–655. Knaap, van W. O. (1986). On the presence of reindeer Rangifer tarandus on Edgeøya, Spitsbergen in the period 3800–5000 B.P. Circumpolar J. 1: 3–10. Leader-Williams, N. (1988). Reindeer on South Georgia. Cambridge Univ. Press, Cambridge. Lister, A. M. (1989). Rapid dwarfing of red deer on Jersey in the last interglacial. Nature 342: 539–542. Lønø, O. (1959). Reinen på Svalbard. Naturen 12: 40–70. [In Norweigian] Maddison, W. P., and Maddison, D. R. (1992). ‘‘MacClade, Version 3.04.’’ Distributed by Sinauer Assosiates, Sunderland, MA. Meldgaard, M. (1986). The Greenland caribou—zoogeography, taxonomy and population dynamics. Meddr. Grønland. Biosci. 20: 1–88.
POLYPHYLETIC ORIGIN OF HIGH-ARCTIC REINDEER Meldgaard, M., and Bennike, O. (1989). Interglacial remains of caribou (Rangifer tarandus) and lemming (Dicrostonyx torquatus(?) ) from North Greenland. Boreas 18: 359–366. Nei, M. (1987). Molecular Evolutionary Genetics. Columbia Univ. Press, New York. Pa¨a¨bo, S., Gifford, J. A., and Wilson, A. C. (1988). Mitochondrial DNA sequences from a 7000-year old brain. Nucleic Acids Res. 16: 9775–9787. Roby, D. D., Thing, H., and Brink, K. L. (1984). History, status and taxanomic identity of caribou (Rangifer tarandus) in Northwest Greenland. Arctic 37: 23–30. Røed, K. (1985). Comparison of the genetic variation in Svalbard and Norwegian reindeer. Can. J. Zool. 63: 2038–2042. Røed, K., Soldal, A. V., and Thorisson, S. (1985). Transferrin variability and founder effect in Icelandic reindeer Rangifer tarandus L. Hereditas 103: 161–164. Røed, K., Staaland, H., Broughton, E., and Thomas, D. C. (1986). Transferrin variation in caribou (Rangifer tarandus L.) on the Canadian Arctic islands. Can. J. Zool. 64: 94–98. Røed, K. H., Mossing, T., Niemen, M., and Rydberg, A. (1987). Transferrin variation and genetic structure of reindeer populations in Scandinavia. Rangifer 7: 12–21.
159
Saccone, C., Pesole, G., and Sbisa´, E. (1991). The main regulatory region of mammalian mitochondrial DNA: Structure–function model and evolutionary pattern. J. Mol. Evol. 33: 83–91. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B., and Erlich, H. A. (1988). Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239: 487–491. Salvigsen, O. (1979). The last deglaciation of Svalbard. Boreas 8: 229–231. Sanger, F., Nichlen, S., and Coulson, A. R. (1977). DNA sequencing with chain-terminal inhibitors. Proc. Natl. Acad. Sci. USA 74: 5463–5468. Swofford, D. L. (1993). ‘‘PAUP Phylogenetic Analysis Using Parsimony Version 3.1,’’ Smithsonian Institution, Washington, DC. Vartanyan, S. L., Garutt, V. E., and Sher, A. V. (1993). Holocene dwarf mammoths from Wrangle Island in the Siberian Arctic. Nature 362: 337–340. Willemsen, G. F. (1983). Osteological measurements and some remarks on the evolution of the Svalbard reindeer Rangifer tarandus platyrhynchus. Z. Sa¨ugetierk. 48: 175–185. Zhurkevich, N. M., and Fomicheva, I. I. (1976). Genetic polymorphism of transferrin of blood serum in reindeer (Rangifer tarandus) indigenous to northeastern Siberia. (In Russian). Genetica 12: 56–65. (not seen).