Content uploaded by Odile Loreille
Author content
All content in this area was uploaded by Odile Loreille on Oct 27, 2017
Content may be subject to copyright.
Short Communication
Molecular phylogeny of the extinct cave lion Panthera leo spelaea
Joachim Burger,
a,*
Wilfried Rosendahl,
b,1
Odile Loreille,
a
Helmut Hemmer,
c,2
Torsten Eriksson,
d,3
Anders G€
otherstr€
om,
e,4
Jennifer Hiller,
f,g,5
Matthew J. Collins,
f,6
Timothy Wess,
g
and Kurt W. Alt
a
a
Molecular Archaeology Mainz, Institute of Anthropology, Johannes Gutenberg-University, Saarstrasse 21, D-55099 Mainz, Germany
b
Institute of Applied Geosciences, Darmstadt University of Technology, Schnittspahnstrasse 9, D-64287 Darmstadt, Germany
c
Institute of Zoology, Johannes Gutenberg-University, Saarstrasse 21, D-55099 Mainz, Germany
d
Bergius Foundation, Royal Swedish Academy of Sciences, Box 50017, SE-10405 Stockholm, Sweden
e
Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Norbyv€
agen 18D, SE-752 36 Uppsala, Sweden
f
School of Civil Engineering and Geoscience, University of Newcastle upon Tyne, Newcastle upon Tyne, UK
g
Biophysics Group, Department of Optometry and Vision Sciences, University of Cardiff, Cardiff, Wales, UK
Received 28 February 2003; revised 25 June 2003
Abstract
To reconstruct the phylogenetic position of the extinct cave lion (Panthera leo spelaea), we sequenced 1 kb of the mitochondrial
cytochrome bgene from two Pleistocene cave lion DNA samples (47 and 32 ky B.P.). Phylogenetic analysis shows that the ancient
sequences form a clade that is most closely related to the extant lions from Africa and Asia; at the same time, cave lions appear to be
highly distinct from their living relatives. Our data show that these cave lion sequences represent lineages that were isolated from
lions in Africa and Asia since their dispersal over Europe about 600 ky B.P., as they are not found among our sample of extant
populations. The cave lion lineages presented here went extinct without mitochondrial descendants on other continents. The high
sequence divergence in the cytochrome bgene between cave and modern lions is notable.
2003 Elsevier Inc. All rights reserved.
1. Introduction
The cave lion [Panthera leo spelaea (Goldfuss, 1810)]
was one of the most important carnivorous competitors
of early man in Europe, from the early Middle Pleisto-
cene onwards. It was an object of Palaeolithic art, such
as the magnificent colour paintings in the Chauvet Cave
(Ard
eche, France) (Lorblanchet, 1995) or the impressive
ivory sculptures from the Vogelherd cave (Swabian Alb,
Germany) (Fig. 1). The first comprehensive morpho-
logical studies of cave lion remains, in the 19th and the
beginning of the 20th century, showed a relationship to
modern lions. Subsequent reinterpretations either linked
cave lions to modern tigers or declared them a separate
species. Osteological revisions have, however, always
indicated a relationship to modern lions (Hemmer,
1974), although lately a new case has been made for a
relationship to tigers based on brain endocasts (Groiss,
1996). Comparative morphological analysis of Pleisto-
cene and Holocene lions on the level of geographic
populations resulted in the description of two basic
evolutionary lines: the spelaea group of the Holarctic
Pleistocene and the leo group of Africa and southern
Asia (Hemmer, 1974). Most authors favour the taxo-
nomic combination of these groups within the species
Panthera leo (Hemmer, 1974; Kurt
en, 1968; Turner and
Ant
on, 1997), but some prefer a taxonomic separation
at the species level, into Panthera spelaea and Panthera
leo (Baryshnikov and Boeskorov, 2001). Here, we report
the mtDNA analysis of two Upper Pleistocene cave
lions, one (Si) 47,180 + 1190/)1040 year B.P. and one
*
Corresponding author. Fax: +49-6131-392-5132.
E-mail addresses: jburger@mail.uni-mainz.de (J. Burger), wilfros@
geo.tu-darmstadt.de (W. Rosendahl), Torsten.Eriksson@bergainska.se
(T. Eriksson), agotherstrom@isciii.es (A. G€
otherstr€
om), j.c.hiller@
stir.ac.uk (J. Hiller), m.collins@ncl.ac.uk (M.J. Collins).
1
Fax: +49-6151-166539.
2
Fax: +49-6136-42424.
3
Fax: +46-08-612-9005.
4
Fax: +46-18-471-63-10.
5
Fax: +44-0-1786-464-994.
6
Fax: +44-0-191-222-5431.
1055-7903/$ - see front matter 2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.ympev.2003.07.020
Molecular Phylogenetics and Evolution 30 (2004) 841–849
MOLECULAR
PHYLOGENETICS
AND
EVOLUTION
www.elsevier.com/locate/ympev
(Ku) 31,890 300 year B.P. old. Our results are con-
sistent with the taxonomy of pantherine cats presented
in Table 1.
2. Materials and methods
2.1. Fossil bone samples
One specimen (Si), an almost complete skeleton of a
cave lion embedded in a greyish silty clay, was exca-
vated in 1985 (Rosendahl and Darga, 2004) at Siegsdorf
in southeastern Bavaria, Germany. For preservation the
bone surfaces were treated with a silica gel resin. Sam-
ples were taken for radiocarbon dating and DNA
analysis from the interior compact bone of the right
femur. The high collagen yield (19.4 wt% bone) suggests
that the bones were not significantly altered diageneti-
cally and that the 14 C-AMS date of 47,180 + 1190/)1040
(KIA 14406) year B.P. is valid. The second, untreated
sample (Ku) was recovered from layer 4 (a light-brown
cave loam) of the Tischhofer cave, 2 km northwest of
Kufstein, Tirol, Austria, 598 m a.s.l, in 1906 (Schlosser,
1910). A pelvis fragment of a juvenile cave lion was
selected for dating and DNA analysis. The collagen
yield was again high (12 wt% bone), suggesting that the
14C-AMS date of 31,890 300 (KIA 16510) year B.P. is
also valid. Modern pantherine samples (P. leo persica
and P. pardus) were from the Frankfurt/Main Zoo,
Germany.
Fig. 1. Cave lion ivory sculpture from Vogelherd cave, Swabian Alb,
Germany (Aurignacian 32 kya).
Table 1
Taxonomy of the pantherine cats (after Hemmer, 1974, 1978, in press, and this study)
Genus Subgenus Species Subspecies
Neofelis Neofelis nebulosa (clouded leopard)
[AJ304497]
Uncia Uncia uncia (snow leopard)
Panthera Tigris Panthera tigris (tiger) [AF053021,
AF053039, AF053048, AF053051]
Panthera Panthera pardus (leopard) [this study]
Panthera onca (jaguar)
Panthera leo (lion) spelaea group (Pleistocene)
Panthera leo fossilis (Early Middle Pleistocene European cave
lion)
Panthera leo vereshchagini (East Siberian and Beringian cave lion)
Panthera leo atrox (North American cave lion)
Panthera leo spelaea (Upper Pleistocene European cave lion) [this
study]
leo group (African and South-Asian lions)
persica subgroup
Panthera leo persica (South Asian lion) [this study]
close to persica subgroup
Panthera leo leo (Atlas lion)
senegalensis subgroup:
Panthera leo senegalensis (West African lion)
Panthera leo azandica (North East Congo lion)
Panthera leo nubica (East African lion) [AF384809, AF384817]
Panthera leo bleyenberghi (Southwest African lion) [AF384811-
AF384815]
Panthera leo krugeri (Southeast African lion) [AF384816,
AF384818]
Panthera leo melanochaita (Cape lion)
Provided as (common name) [sequence origin]. Subspecies nomenclature is given for lions.
842 J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849
2.2. Diagenetic measurements
Fourier-transform infrared spectra were generated
using KBr pellets. The spectra were used to generate a
splitting factor (SF) as described in Termine and Posner
(1966) and Weiner and Bar-Yosef (1990) as well as a
carbonate:phosphate peak ratio (C/P) as shown in
Wright and Schwarcz (1996). These measurements relate
to the degree of mineral alteration in the bone sample.
Powders were also subjected to elemental analysis,
providing a percent value of whole bone nitrogen (% N)
in each sample. This has been determined to relate to the
remaining protein present in archaeological and fossil
samples (Hedges et al., 1995). Finally, powders were
subjected to small-angle X-ray scattering (SAXS) anal-
ysis on a Bruker AXS Nanostar (Karlsruhe) at the
University of Stirling. This provided a detailed mea-
surement of the bone crystallite size and shape in each
sample.
2.3. Ancient DNA work
The DNA work was conducted in two laboratories,
located in separate buildings: one ancient DNA labo-
ratory devoted to pre-PCR procedures and free of other
molecular work, and a second laboratory for the post-
PCR analysis. The extractions were performed in a de-
voted ancient DNA laboratory where no felid DNA had
previously been introduced. All rooms are regularly
washed with bleach and UV-irradiated overnight. Every
item entering these rooms is washed with bleach and
subsequently UV-irradiated. Filtered water for cleaning
is additionally UV-treated for at least 10 h. Two inde-
pendent samples were taken from each specimen for the
Mainz laboratory. A third sample from each lion was
processed in Uppsala to where it was sent directly from
the museums.
2.4. Extraction of ancient DNA samples
0.4–1 g powdered bone samples were incubated in
3–6 mL of extraction buffer (0.5 M EDTA, pH 8.5; 0.5%
N-lauryl sarcosine; 19 mg/mL proteinase K) on a rotary
shaker for 20 h at 37 C. DNA was extracted with phe-
nol/chloroform/isoamyl alcohol (25:24:1); the super-
natant was concentrated by Centricon 30 (Amicon)
dialysis and finally washed several times with UV-trea-
ted HPLC-grade water.
2.5. PCR, cloning, and DNA-sequencing
Twelve PCR products in lengths of 87–209 bp were
designed to cover 1051 bp of the cytochrome bgene of
pantherines (Fig. 2). One additional primer pair ampli-
fies a 474 bp amplicon and is used to test for the presence
of contaminating undegraded DNA in the PCR. When
possible, primers (Table 2) were designed so that they
did not amplify either the human sequence or a known
tiger nuclear insertion. The resulting PCR amplicons
had a minimum of 2 bp and maximum of 59 bp overlap
(without primers). Initially, primers were tested in a
third laboratory so that no molecular work on modern
pantherines was performed in the laboratories prior to
ancient DNA analysis. After aDNA extraction, a series
of PCR amplifications were performed until sequences
from at least two DNA extracts and four independent
PCR runs were available for each amplicon. Further, to
detect possible heterogenous sequences and nuclear
insertions, each PCR product was cloned at least once
and 5–21 clones (average 9) were sequenced. In total, the
Fig. 2. Sequencing strategy. The lengths of the amplicons are shown including primers.The numbers refer to the position on the lion cytochrome b
gene sequence.
J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849 843
sequences were reproduced 14–37 times (average 19) so
that 258 and 191 sequences, respectively, were produced
to establish the 1051 bp sequence of both specimens.
The following sequences were acquired from Gen-
Bank: Felis catus (domestic cat): AB004238; Neofelis
nebulosa (clouded leopard): AJ304497; Panthera leo
(lion): AF384815, AF384811, AF384812, AF384813,
AF384814, AF384818, AF384816, AF384817, and
AF384809; Panthera tigris (tiger): AF053051,
AF053039, AF053048, and AF053021.
The following sequences were produced:
P. leo persica (Asiatic lion) 1 (Zoo Frankfurt a. M.,
Germany); P. leo persica 2 (Zoo Frankfurt a. M., Ger-
many); and P. pardus (leopard) (Zoo Frankfurt a. M.,
Germany).
Amplifications were carried out in a 50 lL reaction
volume containing 50 mM KCl, 10 mM Tris–HCl
(GeneAmp 10Buffer II, PE Applied Biosystems); 2–
2.5 mM MgCl2, 200 lM each dNTPs, 1 lg T4 G32
protein, 0.2 lM of each primer, and 3.5 U of AmpliTaq
Gold (PE Applied Biosystems). The PCR thermal cy-
cling conditions were 94 C for 5 min followed by 38–45
cycles at 94 C for 30 s, at 54–60C for 30 s, and at 72C
for 30 s. One extraction blank and two PCR negative
controls were carried out for each PCR experiment.
The PCR product was purified using the QIAquick
kit from Qiagen. For direct sequencing reactions of both
strands the PRISM kit from Applied Biosystems was
used. Additionally, the PCR products were cloned using
a pUC18 (T-vector, in-house production) transformed
to an Escherichia coli strain (RRI). DNA from selected
clone medium was isolated using the CONCERT Rapid
Plasmid Purification Kit (Gibco, Germany) following
the manufacturerÕs protocol. Five to 21 (average 9)
clones were sequenced using the universal reverse and
forward primers. Sequencing reactions were run on an
Applied Biosystems 310 automatic sequencer.
2.6. Phylogeny
The cytochrome bsequence data set consisted of 20
individuals and 1140 manually aligned positions. Max-
imum likelihood (ML) analysis and bootstrap analyses
used the Linux version of PAUP* (Swofford, 2001) with
the general time-reversible model and gamma distribu-
tion of rates (GTR + G) (Rodr
ıguez et al., 1990; Yang,
1996). The model of evolution was selected by using
MrModeltest, a simplified version of Modeltest 3.06
(Nylander, 2001; Posada and Crandall, 1998). Heuristic
searches in PAUP* used TBR branch swapping on 100
random addition sequence trees, estimating all model
parameters. The bootstrap analysis (Felsenstein, 1985)
was set up to perform 100 replicates with simple addi-
tion sequence of taxa and model parameters set as es-
timated for the best tree found in the ML analysis. The
Bayesian inference analyses used MrBayes (Huelsenbeck
and Ronquist, 2001) and the same model as for ML.
Three separate MrBayes analyses starting from random
trees were performed. In each, 1,000,000 generations of
Markov Chain Monte Carlo (MCMC) were run, sam-
pling a tree every 10 generations. Majority-rule con-
sensus trees were obtained by loading sampled trees into
PAUP* after discarding trees sampled during chain
‘‘burnin’’ (1349, 1749, and 1599 trees were discarded,
respectively). The trees from the three Bayesian analyses
were identical and the posterior probabilities for clades
were almost identical. Their means are used in Fig. 3.
2.7. Divergence time estimates
A likelihood ratio test for rate constancy (Felsenstein,
1988) was performed where the likelihood of our ML tree
was compared to the likelihood of the same tree with the
constraint of a strict molecular clock. The probability for
rejecting the null hypothesis of rate constancy was
0:1>p>0:05 (v210.2632; df ¼18). Since the test did
not reject rate constancy, estimates of divergence times
for nodes were calculated using the clock-based Langley–
Fitch method with the Powell algorithm available in
SandersonÕs r8s (‘‘rates’’) program (Sanderson, 2002).
The most distant outgroup, Felis, was pruned from the
Table 2
Primer sequences (50–30)
CB8u TTTTGAGGTGCAACTGTAATC
CB8l GAAGCCTCCTCAGATTCAC
CB9u TTCCATCCATACTATACAATCAA
CB9l GAGGAGGCGGTTTTCAA
CB10u CTCCGATCTATTCCCAACAAACT
CB10l CCGCTACTAGGAATCAGAATA
CB11u GTGGCCAACCTGTAGAAC
CB11l ATGCCTGAGATGGGTATTA
CB12u CAC ACC CCC TTA TCA AAA TTA TT
CB12l TAA CTG ATG AGA AAG CGG TTA T
CB13u AAATTCTCACCGGCCTCTTTCTA
CB13l TTGGCGTGTAGGTACCGGATAA
CB14u2 CAC ATT TGC CGC GAT GTA AAT
CB14l TGG CCC CAC GGT AAG ACA TAT
CB16u CCA AAC AGC GAG GAA TGA TG
CB16l TTG GCC AAT GGT GAT GAA G
CB17u TAC TAC GGC TCC TAC ACT TTC TCA
CB17l ATG GGA TTG CTG ATA GGA GGT TG
CB19u GATTCTTTGCCTTCCACTTCAT
CB19l2 AAGGCCTAGGATATCTTTGATTGTA
CB20u2 CAGATAAAATTCCATTTCATCCATA
CB20l TGGGGAGGGGTGCTTAGA
CB 21u4 ACC CCG ATA ACT ATA CCC C
CB21l GAGGGCAGGGATAATTGCTAAG
CB22u CATACATTGGGGCCGACCT
CB22l2 ACTACTGCTAGGGCTGAGATGATA
844 J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849
tree prior to the analysis in order to obtain a non-zero
length root branch. The program settings were as fol-
lows: gamma distribution of rates; five time guesses; five
restarts. Confidence intervals (95%) for the estimated
divergence times were also computed (Cutler, 2000) in
r8s with the cutoff parameter set to 2.
An analysis of rate divergence times results in relative
age estimates for all branch points (nodes) in the tree.
In order to convert these to absolute times, it is neces-
sary to fix one node as a calibration point; this point is
therefore, in itself, not estimated by the analysis. We set
the first split of the Panthera leo lineage in our ML tree
to the date of the first appearance of P. leo fossilis in the
European fossil record (marked by an asterisk in Fig. 3).
The earliest date obtained for this appearance is 600 kya
(Garcia Garcia, 2001).
3. Results and discussion
3.1. Diagenetic measurements (Table 3)
To test the general biomolecular state of preservation
of the specimens before starting with the extensive an-
cient DNA analysis, bone samples were subjected to
three separate diagenetic screening techniques: elemental
analysis, Fourier-transform infrared spectroscopy, and
SAXS. Very little mineral alteration is evident in either
archaeological sample, as compared to the modern
values for both SF and C/P. All crystallites measured
using SAXS were determined to be plate-like (data not
shown), which is also consistent with modern unaltered
crystallites (Camacho et al., 1999). The crystallite sizes
of the archaeological samples were significantly lower
than those measured in the modern human sample, but
this smaller size of 2.8–3 nm is consistent with the size
seen in modern faunal samples (Wess and Hiller, un-
published data). In addition, very little nitrogen has
been lost from the archaeological samples, indicating a
high level of residual collagen. The lack of mineral al-
teration and high residual protein, indicate exceptional
bone preservation, suggesting that these bones are suit-
able for recovery of ancient DNA (see Table 3).
3.2. Ancient DNA analysis and authenticity
From each specimen we obtained three separate bone
samples, which were handled and analyzed by three
workers independently in two separate labs (two in
Mainz and one in Uppsala).
The mtDNA sequences derived from each bone un-
derwent multiple verifications using independent sam-
ples, extractions, amplifications, cloning, and sequence
determinations. In all cases, for all three samples the
replicated mtDNA sequences were consistent across all
Table 3
Results of diagenetic screening procedures for cave lion samples Si and Ku, compared to three modern reference samples
Sample SF C/P % N SAXS thicknesses (nm)
Cave lion (Si) 2.76 0.469 3.83 2.99
Cave lion (Ku) 2.91 0.399 3.65 2.98
Modern bear reference 2.82 0.338 4.12 2.93
Modern human reference 2.72 0.445 4.17 3.75
Modern lion reference 2.85 0.375 4.1 2.55
Fig. 3. Maximum likelihood phylogeny of the Panthera clade
(GTR + G model; ln likelihood ¼)3038.43290). The cave lion clade is
in red. Branch lengths are drawn proportional to estimated change;
scale bar 0.01 substitutions per site. Node support values are attached
by the nodes: clade probabilities (Bayesian posterior probabilities in
percent) to the left and bootstrap percentages to the right. The node
used to calibrate divergence time estimates is marked with an asterisk.
Geographical origin of lions is noted within square brackets (P. l.
bleyenberghi,krugeri, and nubica correspond to the senegalensis group;
for subspecies nomenclature, see Table 1). (For interpretation of the
references to colour in this figure legend, the reader is referred to the
web version of this paper.)
J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849 845
experiments. Five nucleotide positions (one in Uppsala
and four in Mainz), however, differed in more than one
PCR from the established consensus sequence. All were
C–T transitions and most likely due to decompositional
deamination events (Gilbert et al., 2003; Hansen et al.,
2001; Hofreiter et al., 2001a). In all cases a number of
additional PCRs was performed, and a vast majority of
sequences confirmed the original cytosine residue.
Contaminations, decompositional base modifica-
tions, nuclear insertions (Zischler et al., 1995), and,
mainly, errors in the procedural design endanger the
interpretation of ancient DNA results. Therefore, for
each specimen the authenticity of the sequence has to be
proven by various criteria (e.g., Hofreiter et al., 2001b).
The authenticity of the sequences presented here is as
ensured as possible, for the following reasons:
•Several different biomolecular screening methods
showed the samples to be exceedingly well-preserved.
•Extraction and PCR blank controls were always
negative.
•Sequences were reproduced various times from at
least two independent extractions, and a total of at
least four independent PCR amplifications.
•Overlapping PCR amplicons always produced the
same sequence.
•In total 3 of the 12 PCR products (125 bp for Si and
163 bp for Ku) including 42 variable positions in Fe-
lidae were replicated in a second lab from a third
bone sample.
•The obtained sequences can both be translated into
an identifiable cytochrome bprotein without non-
sense mutations.
•Attempts to amplify a 474 bp amplicon using panth-
erine specific primers (CB9) failed, indicating
that no modern DNA was involved in the enzymatic
reaction.
•The sequences obtained from two specimens are dif-
ferent from each other, and reproducibly showed this
individual difference.
•Both sequences make sense phylogenetically.
3.3. Cave lion phylogeny
The two 1051 bp cave lion sequences differ from each
other at two base positions. Both are third codon posi-
tion silent substitutions. The South African lion refer-
ence sequence (AF384818) differs from the fossil
sequences by 40/38 silent substitutions and eight substi-
tutions that result in an amino acid change. The cave lion
sequences differ from nine undoubtedly sub-Saharan li-
ons by 47–49 bp (45–47 bp), from two Asian lions (P. leo
persica) by 50/48 bp, from a leopard (P. pardus) by 89/
87 bp, and from four different subspecies of tiger (P. ti-
gris) by 114–117 bp (112–115 bp). These results and the
complete distance matrix in Table 4 agree well with the
overview of pantherine taxonomy presented in Table 1.
We constructed a cladogram from the two cave lion
sequences and extant species of the genus Panthera
(tiger, leopard, and lion; Fig. 3). In accordance with
morphological and behavioural studies of the phyloge-
netic relationships between the extant species of the
genus Panthera, our cytochrome btree shows that the
tiger branch (subgenus Tigris) separated first from
the branch of the jaguar (not shown), the leopard, and
the lion (subgenus Panthera) (cf. Hemmer, 1978). An
earlier analysis of 358 bp of the mitochondrial 12S RNA
coding DNA and 289 bp of the cytochrome bgene is
consistent with these relationships (Janczewski et al.,
1995). The leopard and the lion represent the last species
separation within the jaguar, leopard, and lion clade
(Janczweski et al., 1995; Peters and Tonkin-Leyhausen,
1999). According to the palaeontological record, the first
divergence in the subgenus Panthera took place in the
late Middle Villafranchian at the end of the Pliocene,
about 1900 kya, with the dispersal of the stem species
out of Africa. This gave rise to the Holarctic base jaguar
population (Hemmer et al., 2001). Therefore, the split
between the subgeneric Tigris and Panthera clades can-
not have been a later event, but rather an earlier one.
Unfortunately, well-founded palaeontological dating is
not yet possible for this point (Hemmer et al., 2001).
Evolutionary rate constancy was not rejected for our
data, and clock-based estimates of divergence times were
therefore obtained. The age of the split between the
subgeneric Tigris and Panthera clades was estimated to
1428–2295 kya (95% confidence interval), and the leo-
pard–lion split to 1000–1559 kya. The latter split has
not been unequivocally dated with palaeontological
data, but our estimate is consistent with a likely Upper
Villafranchian event at the beginning of the Lower
Pleistocene.
Phylogenetic divergence within lions is marked by
their dispersal over Europe in the early Middle Pleisto-
cene, not before the Cromerian interglacial III or IV
(Garcia Garcia, 2001), i.e., not before 600 kya. From
this time on, the cave lion (spelaea group) developed in
Europe, beginning with the early Middle Pleistocene
Panthera leo fossilis and ending with the Upper Pleis-
tocene P. leo spelaea (Hemmer, 1974). Our estimate for
the more recent divergence within the leo group into the
African and Asian extant lion subgroups, the senegal-
ensis group in Africa (comprising all sub-Saharan Afri-
can lions; Hemmer, 1974, in press) and persica in Asia
(the north African Barbary lion, leo, is closer to the
latter), is 74–203 kya. This split has not been dated be-
fore using palaeontological data.
In the context of this study, we have shown that a
considerable mitochondrial genetic distance exists be-
tween these two cave lions and extant lions, one that is
much larger than the range of genetic variation seen in
living populations of lions. These results imply that our
cave lion lineages were genetically isolated from the
846 J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849
Table 4
Distance matrix of sequences used in this study
1 2 3 4 5 6 7 8 9 1011121314151617181920
1Felis —
2Neofelis 0.3224 —
3P. leo (Botswana) 0.30831 0.28851 —
4P. leo (Namibia) 0.33078 0.28764 0.00104 —
5P. leo (Namibia) 0.30831 0.28851 0.00000 0.00104 —
6P. leo (Namibia) 0.30831 0.28851 0.00000 0.00104 0.00000 —
7P. leo (Namibia) 0.30831 0.28851 0.00000 0.00104 0.00000 0.00000 —
8P. leo (Natal) 0.32005 0.27911 0.00731 0.00430 0.00731 0.00731 0.00731 —
9P. leo (Transvaal) 0.32005 0.27911 0.00731 0.00430 0.00731 0.00731 0.00731 0.00000 —
10 P. leo (Kenya) 0.32005 0.27911 0.00731 0.00430 0.00731 0.00731 0.00731 0.00000 0.00000 —
11 P. leo (Uganda) 0.31044 0.27911 0.00635 0.00321 0.00635 0.00635 0.00635 0.00451 0.00451 0.00451 —
12 P. leo persica 0.35607 0.28293 0.01220 0.01164 0.01220 0.01220 0.01220 0.01004 0.01004 0.01004 0.00900 —
13 P. leo persica 0.35607 0.28293 0.01220 0.01164 0.01220 0.01220 0.01220 0.01004 0.01004 0.01004 0.00900 0.00000 —
14 P. leo spelaea 0.36026 0.26236 0.05904 0.06103 0.05904 0.05904 0.05904 0.05896 0.05896 0.05896 0.06043 0.06159 0.06159 —
15 P. leo spelaea 0.35443 0.26206 0.05620 0.05789 0.05620 0.05620 0.05620 0.05618 0.05618 0.05618 0.05760 0.05893 0.05893 0.00193 —
16 P. pardus 0.26605 0.36516 0.13970 0.13844 0.13970 0.13970 0.13970 0.13945 0.13945 0.13945 0.13701 0.13818 0.13818 0.13696 0.13299 —
17 P. tigris corbetti 0.29330 0.26427 0.19150 0.19435 0.19150 0.19150 0.19150 0.20602 0.20602 0.20602 0.20063 0.22154 0.22154 0.22123 0.21651 0.20798 —
18 P. tigris altai 0.28807 0.27977 0.18806 0.19003 0.18806 0.18806 0.18806 0.20228 0.20228 0.20228 0.19704 0.21732 0.21732 0.21726 0.21256 0.20423 0.00088 —
19 P. tigris sumatrae 0.27307 0.28704 0.18106 0.18410 0.18106 0.18106 0.18106 0.19481 0.19481 0.19481 0.18977 0.20889 0.20889 0.21445 0.20967 0.20164 0.00633 0.00541 —
20 P. tigris tigris 0.27523 0.26569 0.18258 0.18598 0.18258 0.18258 0.18258 0.18938 0.18938 0.18938 0.18451 0.20268 0.20268 0.20851 0.20376 0.19601 0.00451 0.00358 0.00358 —
See Table 1 for GenBank accession numbers. Distances are corrected using the same model of evolution as in the maximum likelihood analysis. A general time-reversible model was used, with
rates assumed to follow gamma distribution with a shape parameter ¼0.244; this was estimated for the best maximum likelihood tree found. Numbers 3–7 are Panthera leo bleyenberghi, 8 and 9 are
P. leo krugeri, and 10 and 11 are P. leo nubica.
J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849 847
ancestors of modern Asian and African lions from the
early Middle Pleistocene onwards, and went extinct
without contributing mitochondrial DNA to extant
lineages. The question of whether the cave lion popu-
lation of Europe and the extant lion populations of
Africa and Asia should be recognized as different species
may be a matter of convention. In our tree, the cave lion
clade is a sister to the extant lions, which means that the
cave lions may be excluded from or included within the
species P. leo. However, the maximum possible number
of lion generations since the 600 ky split in comparison
with other pantherine species argues for the single spe-
cies nomenclature (Hemmer, in press).
If it is assumed that the two cave lion sequences are
representative for the European cave lions of that time,
the very good support for both this clade (100% boot-
strap as well as posterior probability) and for the clade
of extant lions indicates that European cave lion popu-
lations may have left no mitochondrial descendants,
whereas the mitochondrial genes of their contempora-
neous African and Asian relatives survive in extant lion
populations. It remains to be seen if the considerable
sequence divergence between the clades (nearly 5%) will
remain if the sample of fossil and extant specimens is
increased. Further studies are needed to show if genetic
changes or characteristics, as well as ecological factors,
may have played a role in the extinction of cave lion
populations in Europe at the end of the Pleistocene. This
study represents another successful use of modern ge-
netic and phylogenetic techniques to investigate long-
standing palaeontological questions, and it is our hope
that such studies continue to shed light on issues of
evolutionary descent and speciation that cannot be as
well elucidated by other means.
Acknowledgments
We thank Robert Darga, Naturkunde- und Mam-
mutmuseum Siegsdorf, and Hugo Oberkofler, Museum
Kufstein for providing cave lion samples. We are in-
debted to Sabine Hilsberg and Bert Geyer from the Zoo
in Frankfurt/Main for providing blood samples, and
Sabine Moeller-Rieker and Ursula Arndt for assistance.
J.H. gratefully acknowledges funding from the Well-
come Trust Programme in Bioarchaeology. We thank
Prof. Nicholas Conard for providing Fig. 1.
References
Baryshnikov, G.F., Boeskorov, G., 2001. The Pleistocene cave lion,
Panthera spelaea (Carnivora, Felidae) from Yakutia, Russia.
Cranium 18, 7–24.
Camacho, N.P., Rinnerthaler, S., Paschalis, E.P., Mendelsohn, R.,
Boskey, A.L., Fratzl, P., 1999. Complementary information on
bone ultrastructure from scanning small angle X-ray scattering and
Fourier-transform infrared microspectroscopy. Bone 25, 287–293.
Cutler, D.J., 2000. Estimating divergence times in the presence of an
overdispersed molecular clock. Mol. Biol. Evol. 17, 1647–1660.
Felsenstein, J., 1985. Confidence limits on phylogenies: an approach
using bootstrap. Evolution 39, 783–791.
Felsenstein, J., 1988. Phylogenies from molecular sequences: inference
and reliability. Annu. Rev. Genet. 22, 521–565.
Garcia Garcia, N., 2001. Los Carnivorors de los Yacimientos
Pleistocenos de la Sierra de Atapuerca. Thesis Doctoral.
Gilbert, M.T., Hansen, A.J., Willerslev, E., Rudbeck, L., Barnes, I.,
Lynnerup, N., Cooper, A., 2003. Characterization of genetic
miscoding lesions caused by postmortem damage. Am. J. Hum.
Genet. 72, 48–61.
Groiss, J.Th., 1996. Der H€
ohlentiger Panthera tigris spelaea (Gold-
fuss). Neves Jahrb. Geol. Pal€
aont. Abh. 7, 399–414.
Hansen, A.J., Willerslev, E., Wiuf, C., Mourier, T., Arctander, P.,
2001. Statistical evidence for miscoding lesions in ancient DNA
templates. Mol. Biol. Evol. 18, 262–265.
Hedges, R.E.M., Millard, A.R., Pike, A.W.G., 1995. Measurements
and relationships of diagenetic alteration of bone from three
archaeological sites. J. Archaeol. Sci. 22, 201–209.
Hemmer, H., 1974. Untersuchungen zur Stammesgeschichte der
Pantherkatzen (Pantherinae), Teil III: Zur Artgeschichte des
L€
owen Panthera (Panthera) leo (Linnaeus 1758). Ver€
offentlichun-
gen der Zoologischen Staatssammlung M€
unchen 17, 167–280.
Hemmer, H., 1978. The evolutionary systematics of living Felidae:
present status and current problems. Carnivore 1, 71–79.
Hemmer, H., in press. Pleistoz€
ane Katzen Europas—eine €
Ubersicht.
Cranium.
Hemmer, H., Kahlke, R.-D., Vekua, A.K., 2001. The Jaguar-Panthera
onca gombaszoegensis (Kretzoi, 1938) (Carnivora: Felidae) in the
late Lower Pleistocene of Akhalkalaki (South Georgie; Transcauc-
asia) and its evolutionary and ecological significance. Geobios 34,
475–486.
Hofreiter, M., Jaenicke, V., Serre, D., Haeseler, A.A., P€
a€
abo, S.,
2001a. DNA sequences from multiple amplifications reveal arti-
facts induced by cytosine deamination in ancient DNA. Nucleic
Acids Res. 29, 4793–4799.
Hofreiter, M., Serre, D., Poinar, H.N., Kuch, M., P€
a€
abo, S., 2001b.
Ancient DNA. Nat. Rev. Genet. 2, 353–359.
Huelsenbeck, J.P., Ronquist, F., 2001. A program for the Bayesian
inference phylogeny, version 2.01.
Janczewski, D.N., Modi, W.S., Stephens, J.C., OÕBrien, S.J., 1995.
Molecular evolution of mitochondrial 12S RNA and cytochrome b
sequences in the pantherine lineage of Felidae. Mol. Biol. Evol. 12,
690–707.
Kurt
en, B., 1968. Pleistocene Mammals of Europe. Weidenfeld and
Nicolson, London.
Lorblanchet, M., 1995. Les Grottes Orn
ees de la Prehistoire. Errance,
Paris.
Nylander, J.A.A., 2001. MrModeltest, Computer program, version
1.1b. Uppsala University.
Peters, G., Tonkin-Leyhausen, B.A., 1999. Evolution of acoustic
communication signals of mammals: friendly close-range vocaliza-
tions in Felidae (Carnivora). J. Mamm. Evol. 6, 129–159.
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of
DNA substitution. Bioinformatics 14, 817–818.
Rodr
ıguez, F., Oliver, J.L., Marin, A., Medina, J.R., 1990. The general
stochastic model of nucleotide substitution. J. Theor. Biol. 142,
485–501.
Rosendahl, W., Darga, R., 2004. Homo sapiens neanderthalensis
et Panthera leo spelaea—du nouveau
a propos du site de Siegsdorf
(Chiemgau), Bavi
ere/Allemagne. Rev. Palaeobiol. 3, 2.
Sanderson, M.J., 2002. Analysis of rates (r8s) of evolution (and other
stuff), Computer program, version 1.5. University of California,
Davis.
848 J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849
Schlosser, M., 1910. Die B€
aren-oder Tischoferh€
ohle im Kaisertal
bei Kufstein. Abh. Math-physik. K€
onigl. Bayr. Akad. Wiss. 24,
385–506.
Swofford, D.L., 2001. PAUP*. Phylogenetic analysis using parsimony
and other methods, version 4.0.
Termine, J.D., Posner, A.S., 1966. Infra-red determination of the
percentage of crystallinity in apatitic calcium phosphates. Nature
211, 268–270.
Turner, A., Ant
on, M., 1997. The Big Cats and their Fossil Relatives.
Columbia University Press, New York.
Weiner, S., Bar-Yosef, O., 1990. States of preservation of bones from
prehistoric sites in the Near East: a survey. J. Archaeol. Sci. 17, 196.
Wright, L.E., Schwarcz, H.P., 1996. Infrared and isotopic evidence for
diagenesis of bone apatite at Dos Pilas Guatamala: palaeodietary
implications. J. Archaeol. Sci. 23, 933–944.
Yang, Z., 1996. Among-site variation and its impact on phylogenetic
analyses. TREE 11, 367–371.
Zischler, H., H€
oss, M., Handt, O., von Haeseler, A., P€
a€
abo, S.,
Goudsimt, J., van der Kuyl, A.C., 1995. Detecting dinosaur DNA.
Science 268, 1192–1193.
J. Burger et al. / Molecular Phylogenetics and Evolution 30 (2004) 841–849 849
