A genomic approach to examine the complex evolution of laurasiatherian mammals.
ABSTRACT Recent phylogenomic studies have failed to conclusively resolve certain branches of the placental mammalian tree, despite the evolutionary analysis of genomic data from 32 species. Previous analyses of single genes and retroposon insertion data yielded support for different phylogenetic scenarios for the most basal divergences. The results indicated that some mammalian divergences were best interpreted not as a single bifurcating tree, but as an evolutionary network. In these studies the relationships among some orders of the super-clade Laurasiatheria were poorly supported, albeit not studied in detail. Therefore, 4775 protein-coding genes (6,196,263 nucleotides) were collected and aligned in order to analyze the evolution of this clade. Additionally, over 200,000 introns were screened in silico, resulting in 32 phylogenetically informative long interspersed nuclear elements (LINE) insertion events. The present study shows that the genome evolution of Laurasiatheria may best be understood as an evolutionary network. Thus, contrary to the common expectation to resolve major evolutionary events as a bifurcating tree, genome analyses unveil complex speciation processes even in deep mammalian divergences. We exemplify this on a subset of 1159 suitable genes that have individual histories, most likely due to incomplete lineage sorting or introgression, processes that can make the genealogy of mammalian genomes complex. These unexpected results have major implications for the understanding of evolution in general, because the evolution of even some higher level taxa such as mammalian orders may sometimes not be interpreted as a simple bifurcating pattern.
Article: Resolution among major placental mammal interordinal relationships with genome data imply that speciation influenced their earliest radiations.[show abstract] [hide abstract]
ABSTRACT: A number of the deeper divergences in the placental mammal tree are still inconclusively resolved despite extensive phylogenomic analyses. A recent analysis of 200 kbp of protein coding sequences yielded only limited support for the relationships among Laurasiatheria (cow, dog, bat and shrew), probably because the divergences occurred only within a few million years from each other. It is generally expected that increasing the amount of data and improving the taxon sampling enhance the resolution of narrow divergences. Therefore these and other difficult splits were examined by phylogenomic analysis of the hitherto largest sequence alignment. The increasingly complete genome data of placental mammals also allowed developing a novel and stringent data search method. The rigorous data handling, recursive BLAST, successfully removed the sequences from gene families, including those from well-known families hemoglobin, olfactory, myosin and HOX genes, thus avoiding alignment of possibly paralogous sequences. The current phylogenomic analysis of 3,012 genes (2,844,615 nucleotides) from a total of 22 species yielded statistically significant support for most relationships. While some major clades were confirmed using genomic sequence data, the placement of the treeshrew, bat and the relationship between Boreoeutheria, Xenarthra and Afrotheria remained problematic to resolve despite the size of the alignment. Phylogenomic analysis of divergence times dated the basal placental mammal splits at 95-100 million years ago. Many of the following divergences occurred only a few (2-4) million years later. Relationships with narrow divergence time intervals received unexpectedly limited support even from the phylogenomic analyses. The narrow temporal window within which some placental divergences took place suggests that inconsistencies and limited resolution of the mammalian tree may have their natural explanation in speciation processes such as lineage sorting, introgression from species hybridization or hybrid speciation. These processes obscure phylogenetic analysis, making some parts of the tree difficult to resolve even with genome data.BMC Evolutionary Biology 02/2008; 8:162. · 3.52 Impact Factor
[show abstract] [hide abstract]
ABSTRACT: The massive amount of genomic sequence data that is now available for analyzing evolutionary relationships among 31 placental mammals reduces the stochastic error in phylogenetic analyses to virtually zero. One would expect that this would make it possible to finally resolve controversial branches in the placental mammalian tree. We analyzed a 2,863,797 nucleotide-long alignment (3,364 genes) from 31 placental mammals for reconstructing their evolution. Most placental mammalian relationships were resolved, and a consensus of their evolution is emerging. However, certain branches remain difficult or virtually impossible to resolve. These branches are characterized by short divergence times in the order of 1-4 million years. Computer simulations based on parameters from the real data show that as little as about 12,500 amino acid sites could be sufficient to confidently resolve short branches as old as about 90 million years ago (Ma). Thus, the amount of sequence data should no longer be a limiting factor in resolving the relationships among placental mammals. The timing of the early radiation of placental mammals coincides with a period of climate warming some 100-80 Ma and with continental fragmentation. These global processes may have triggered the rapid diversification of placental mammals. However, the rapid radiations of certain mammalian groups complicate phylogenetic analyses, possibly due to incomplete lineage sorting and introgression. These speciation-related processes led to a mosaic genome and conflicting phylogenetic signals. Split network methods are ideal for visualizing these problematic branches and can therefore depict data conflict and possibly the true evolutionary history better than strictly bifurcating trees. Given the timing of tectonics, of placental mammalian divergences, and the fossil record, a Laurasian rather than Gondwanan origin of placental mammals seems the most parsimonious explanation.Molecular Biology and Evolution 12/2010; 27(12):2804-16. · 5.55 Impact Factor
Article: Gene Trees in Species Trees[show abstract] [hide abstract]
ABSTRACT: Exploration of the relationship between gene trees and their containing species trees leads to consideration of how to reconstruct species trees from gene trees and of the concept of phylogeny as a cloud of gene histories. When gene copies are sampled from various species, the gene tree relating these copies might disagree with the species phylogeny. This discord can arise from horizontal transfer (including hybridization), lineage sorting, and gene duplication and extinction. Lineage sorting could also be called deep coalescence , the failure of ancestral copies to coalesce (looking backwards in time) into a common ancestral copy until deeper than previous speciation events. These events depend on various factors; for instance, deep coalescence is more likely if the branches of the species tree are short (in generations) and wide (in population size). A similar dependence on process is found in historical biogeography and host-parasite relationships. Each of the processes of discord could yield a different parsimony criterion for reconstructing the species tree from a set of gene trees: with horizontal transfer, choose the species tree that minimizes the number of transfer events; with deep coalescence, choose the tree minimizing the number of extra gene lineages that had to coexist along species lineages; with gene duplication, choose the tree minimizing duplication and/or extinction events. Maximum likelihood methods for reconstructing the species tree are also possible because coalescence theory provides the probability that a particular gene tree would occur given a species tree (with branch lengths and widths specified). In considering these issues, one is provoked to reconsider precisely what is phylogeny. Perhaps it is misleading to view some gene trees as agreeing and other gene trees as disagreeing with the species tree; rather, all of the gene trees are part of the species tree, which can be visualized like a fuzzy statistical distribution, a cloud of gene histories. Alternatively, phylogeny might be (and has been) viewed not as a history of what happened, genetically, but as a history of what could have happened, i.e., a history of changes in the probabilities of inter-breeding.
A Genomic Approach to Examine the Complex Evolution
of Laurasiatherian Mammals
Bjo ¨rn M. Hallstro ¨m1*, Adrian Schneider2, Stefan Zoller3, Axel Janke1,4
1Biodiversity and Climate Research Centre (BiK-F) & Senckenberg Gesellschaft fu ¨r Naturforschung, Frankfurt am Main, Germany, 2University of Edinburgh, Institute of
Evolutionary Biology, Edinburgh, United Kingdom, 3ETH Zurich, Computational Biochemistry Research Group, Zurich, Switzerland, 4Goethe University, Institute for
Ecology, Evolution and Diversity, Frankfurt am Main, Germany
Recent phylogenomic studies have failed to conclusively resolve certain branches of the placental mammalian tree, despite
the evolutionary analysis of genomic data from 32 species. Previous analyses of single genes and retroposon insertion data
yielded support for different phylogenetic scenarios for the most basal divergences. The results indicated that some
mammalian divergences were best interpreted not as a single bifurcating tree, but as an evolutionary network. In these
studies the relationships among some orders of the super-clade Laurasiatheria were poorly supported, albeit not studied in
detail. Therefore, 4775 protein-coding genes (6,196,263 nucleotides) were collected and aligned in order to analyze the
evolution of this clade. Additionally, over 200,000 introns were screened in silico, resulting in 32 phylogenetically informative
long interspersed nuclear elements (LINE) insertion events. The present study shows that the genome evolution of
Laurasiatheria may best be understood as an evolutionary network. Thus, contrary to the common expectation to resolve
major evolutionary events as a bifurcating tree, genome analyses unveil complex speciation processes even in deep
mammalian divergences. We exemplify this on a subset of 1159 suitable genes that have individual histories, most likely due
to incomplete lineage sorting or introgression, processes that can make the genealogy of mammalian genomes
complex. These unexpected results have major implications for the understanding of evolution in general, because the
evolution of even some higher level taxa such as mammalian orders may sometimes not be interpreted as a simple
Citation: Hallstro ¨m BM, Schneider A, Zoller S, Janke A (2011) A Genomic Approach to Examine the Complex Evolution of Laurasiatherian Mammals. PLoS
ONE 6(12): e28199. doi:10.1371/journal.pone.0028199
Editor: Ed Louis, University of Nottingham, United Kingdom
Received August 4, 2011; Accepted November 3, 2011; Published December 2, 2011
Copyright: ? 2011 Hallstro ¨m et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The present study was funded by the research funding programme ‘‘ LOEWE - Landes -Offensive zur Entwicklung Wissenschaftlich-o ¨konomischer
Exzellenz’’ of Hesse’s Ministry of Higher Education, Research, and the Arts. The funders had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
While the placental mammalian tree is becoming increasingly
better resolved, it has proven difficult to fully resolve several
branches of it as a bifurcating tree, despite the availability and
analyses of whole genome data [1,2]. While the sheer amount of
genomic data should be sufficient to resolve very short branches
within the placental mammalian tree , the support for some
branches is often ambigious. Interestingly, these problematic
branches are characterized by rapid divergences within 1–3 million
years (Myr) . This makes it possible that speciation related
processes, such as incomplete lineage sorting or introgression, lead
to gene trees that differ from the species tree [3,4]. The complex
pattern of retroposon insertion data for the earliest placental
mammalian divergences  corroborate this idea, suggesting that a
network-like evolution instead of a bifurcating tree best depict and
interpret the evolutionary process . Other such problematic
relationships among placental mammals have been identified by
phylogenomic [1,2] and retroposon insertion data . A case in
point isthe evolutionofthe mammaliancladeLaurasiatheria,which
comprises several orders of placental mammals.
Laurasiatheria include the classical orders Perissodactyla,
Carnivora, Pholidota, Artiodactyla, Cetacea, Chiroptera, and
Lipotyphla [7,8]. Initially, morphological and early molecular
studies spread these orders to different parts of the mammalian
tree or left their position unresolved . More detailed molecular
phylogenetic studies grouped these diverse orders into one clade,
Laurasiatheria. Early mitogenomic studies suggested a close
relationship between carnivores and perissodactyls and this group
in turn joined Cetartiodactyla [10,11]. Later mitogenomic studies
added Chiroptera  and parts of a then paraphyletic Lipotyphla
to the Laurasiatheria clade , whereas analysis of nuclear genes
placed all of Lipotyphla within Laurasiatheria . Finally, the
Pholidota (pangolin) were joined with the carnivores by nuclear
and mitogenomic studies [11,14].
Currently molecular phylogenetic studies generally agree on a
(Chiroptera,(Cetartiodactyla, (Perissodactyla, (Pholidota, Carnivo-
ra))) branching order [11,14]. With the exception of the Pholidota,
which lack large-scale genomic sequence data, recent phyloge-
nomic analyses generally support this topology. However, the
relationships remained only poorly supported despite the use of
some 3 million nucleotides of sequence data from 3400 protein
coding genes .
So far only one study using rare genomic events such as data
from retroposon insertions has been made to study the
relationships within Laurasiatheria. In contrast to sequence-based
PLoS ONE | www.plosone.org1 December 2011 | Volume 6 | Issue 12 | e28199
studies, analyses of three retroposon insertions support the
grouping of Chiroptera with the Perissodactyla/Carnivora and a
new name for this unexpected clade – Pegasoferae – has been
suggested . Yet, this study found one retroposon insertion event
that contradicted the Chiroptera plus Perissodactyla/Carnivora
grouping. This one retroposon insertion supports the traditional
sequence-analyses based placement of Chiroptera.
A recent phylogenomic study of mammalian relationships
involved all tetrapod species from which whole genome data were
available. While it is advantageous to increase taxon sampling, this
approach leads to the exclusion of large amounts of sequence data
when stringent data collection and alignment strategies are
employed [1,2]. In addition, the inclusion of distantly related
species in the analyses even make it possible that orthologs are
misidentified, and thus excluded, as paralogs by overly stringent
data retrieval algorithms such as recursive BLAST .
In order to specifically analyse laurasiatherian relationships with
a dataset maximized for the amount of phylogenetically informa-
tive data, only human and mouse are used as outgroups to root the
tree in this study. These species have among the best genomic
sequence coverage and annotation. Furthermore, there is an
unequivocal consensus that these two species are joined in the
clade Euarchontoglires which is the sister group to Laurasiatheria
within Boreotheria. Thus, human and mouse are the ideal
outgroups for this study. By also utilizing the recently released
genome of the giant panda (Ailuropoda melanoleuca) , this
approach allows the collection of a larger number of genes from
more species than in previous phylogenomic studies. Therefore,
analyses based on concatenated data and single genes allow for a
more detailed study of laurasiatherian relationships. In addition,
the quality and quantity of the genome data have been steadily
improving. This makes in silico searches for phylogenetic
informative retroposon insertion data feasible for evaluating
hypotheses that were based on sequence data analysis. Long
interspersed nuclear elements 1 (LINE 1) retroposon sequences
were used for these searches, because these elements were active
during this time of placental mammalian evolution and have
successfully been used in other phylogenetic studies [16–18].
These are currently the only known retroposons that are common
to different orders, while short interspersed nuclear elements
(SINEs) are order-specific .
The complementary DNA (cDNA) databases for all species
included in the study, except the panda, were downloaded from
Ensembl (http://www.ensembl.org, release 57). The whole
genome sequence of the panda was downloaded from the Giant
Panda Database (http://panda.genomics.org.cn/) and the cDNA
sequences were extracted using the gene annotation based on
homology to dog genes. Table 1 lists the included species. For
some comparisons the genome data from the opossum were
included in the analyses.
Data collection and alignment was, with a few exceptions,
performed as described previously  and is thus only briefly
detailed here. Orthologs were identified with the recursive BLAST
method . Sequences were translated to amino acids and aligned
using MUSCLE . The resulting alignments were then back-
translated to nucleotides. Any alignment showing an overall
nucleotide difference larger than 30% between any two species
was discarded. As an additional filtering step, uninformative
quickly evolving sites were eliminated by the program Noisy,
version 1.5.9 .
Phylogenetic analysis using maximum likelihood was performed
using the programs Treefinder (TF)  and RAxML 7.0.4 ,
applying the GTR model  to nucleotide data and WAG2000
 to amino acid data. In both cases, rate heterogeneity was
applied using 4 gamma rate categories, 4G+I. Both heuristic
searches and exhaustive tree comparisons, under the assumption
of monophyletic orders were performed. Divergence times were
estimated from overall best amino acid (AA) ML tree using 6
calibration points  (Table S1) and the nonparametric rate
smoothing method on a logarithmic scale (NPRS-LOG) as
implemented in TF .
Codon-based tree reconstruction was performed using the
Markov Chain Monte Carlo (MCMC) method implemented in
BEAST , using its BEAGLE library for computing on graphics
processing units (GPUs). This decreases computation times by a
factor of up to 90 . The analyses were performed using a semi-
parametric codon model based on principal component analysis of
mammalian sequence data . For each alignment and topology,
the model-parameters as well as the branch lengths were
optimized with a chain length of 700,000 sampled every 500 tree.
Instead of maximizing the likelihood, BEAST allows an estimation
of the marginal log-likelihood (mLogL) by integrating over the
whole parameter space [30,31]. Tree and model comparisons can
be performed using the Bayes factor , which can be
approximated as the difference of the mLogLs. For 229 trees,
BEAST failed to successfully optimize the parameters. These trees
were excluded from further analysis.
In addition to the analyses of concatenated data, all gene
alignments with sequence data from all species were analyzed
separately. The problem was reduced to resolving the relationships
of four orders, leaving 15 possible topologies that were individually
evaluated by ML analyses for each of the 1159 gene alignments.
The same models as outlines above were used with parameters
estimated from individual alignments. Information from likelihood
maximizations on the 1561159 gene trees were analyzed by
Table 1. The names, order and sequence coverage of the
species included in this study.
Common nameBinomial nameOrderCoverage
DogCanis familiaris Carnivora100%
Cat Felis catusCarnivora67.7%
Giant PandaAlluropoda melanoleucaCarnivora98.1%
Horse Equus caballus Perissodactyla98.0%
CowBos taurus Cetartiodactyla92.9%
Bottlenose Dolphin Tursiops truncatus Cetartiodactyla91.8%
PigSus domestica Cetartiodactyla77.7%
AlpacaVicugna pacos Cetartiodactyla66.7%
Large Flying FoxPteropus vampyrusChiroptera 91.7%
Little Brown Bat Myotis lucifugusChiroptera77.1%
European Hedgehog Erinaceus europaeus Erinaceomorpha 74.8%
Common ShrewSorex araneusSoricomorpha68.1%
Human Homo sapiensPrimates100%
House MouseMus musculusRodentia90.8%
Note – Coverage give the percent sequence coverage in the 6,196,263 nt
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org2December 2011 | Volume 6 | Issue 12 | e28199
counting how often each topology was among the most likely trees
and how often a topology was rejected by another topology with a
significantly higher likelihood. A significantly higher likelihood is
defined as one that is larger than two log-likelihood units from the
original. When different topologies had the same likelihood for
single-gene alignments, they were counted individually for each
tree. In addition, all likelihood values for a given topology and data
set were added up in order to compare the total likelihoods of the
different topologies. This approach corresponds to the ‘‘separate’’
analysis according to the definition in Pupko et al., . Since the
mLogLs of the codon-based analysis are expected values and not
maxima, so typically no two topologies end up with exactly the
same mLogL. Thus, a tolerance of 0.5 LogL units was used, and
two values that lay within two mLogLs of each other were
considered as being equal. A topology was rejected if its mLogL
was 10 units lower than the highest. The ML trees from the single-
gene analysis were also used to construct a consensus network
using the SplitsTree4 program , which is used to illustrate the
conflicts of the phylogenetic signal.
For the five most likely tree topologies, the influence of several
properties of the sequences on the outcome of the codon analysis
was tested. The evaluated factors were alignment length, longest
distance among the 15 sequences, sum of all pair wise distances,
deviation of the codon usage frequencies from the average over all
alignments and deviation of the nucleotide usage frequencies from
the average. For each factor, the alignments were divided into two
equal-sized groups; those with the largest values and those with the
smallest values. It was then counted how often each topology was
the only one with highest mLogL. Chi-square tests were performed
to quantify the significance of the difference between the two
‘‘best’’ distributions of topologies.
Finally, a multilocus Bayesian analysis using the program BEST
 was performed. This method attempts to construct a species
tree using a multiple estimated gene trees. This is done by utilizing
a Bayesian hierarchical model to combine traditional phyloge-
netics with coalescent theory. 763 genes (1,313,880 nucleotide
characters) were selected for maximum alignment coverage and
length. This data set was analyzed in BEST, with all parameters
unlinked, runnning for 15,000,000 generations, with two simul-
taneous runs each with one ‘‘heated’’ and one ‘‘cold chain. The
first 1.500.000 generations (10%) were discarded as burnin.
For the retroposon insertion analysis intron sequences longer
than 300 bp and shorter than 3000 bp were collected from the
Ensembl database (version 49) for the cow, dog, horse and
microbat genomes, respectively. Between 40,000 and 95,000
introns were identified in each of the species above. Retroposed
elements in these introns were identified using the program
RepeatMasker version 3.2.8 (http://www.repeatmasker.org/).
From all identified repeated elements only LINE1 elements were
considered for the search and phylogenetic analysis. In total
47,535 LINE1 elements were identified, of which 22,873 were
found in the horse genome, 13,359 in the dog genome, 6,557 in
the cow genome and 4,756 in the bat genome. Using these intron
sequences the orthologous region in the other three species were
identified. The full sequences of orthologous genes were extracted,
based on Ensembl orthology data. The relevant intron sequences
were located by making local pair wise alignments with 80 bp of
exon sequence located upstream and downstream of the intron. In
cases where the intron could be located in all four species a four-
way multiple sequence alignment was created using MAFFT .
This resulted in 19,725 alignments that were guided by 7576
retropson insertions that were initially identified in the horse, 7248
in the dog , 2793 in the cow, and 2108 in the bat, respectively. All
four-way alignments were screened for retroposons that were
present in either two or three of the species, and absent in the
others. These retroposon inserts were considered potential markers
for the phylogenetic relationships between the four orders. Finally
intron sequences from the remaining laurasiatherian species and,
when possible, outgroup sequences were added to the alignments.
Additionally, several hundred alignments in which the insertion
was present in either only one or all four species were randomly
selected and manually screened for potentially informative retro-
poson markers for other parts of the laurasiatherian tree. The
alignments of the, in total 25, informative L1 retroposon insertions
that were used for the tree and network are shown in Figure S2. A
consensus tree, was constructed with SplitsTree4 from all partial
trees corresponding to the L1 retroposon data for the conflicting
hypotheses among the four laurasiatherian orders. The branch
lengths are determined by the number of retroposon insertions
supporting each topology.
The final alignment consisted of 6,196,263 nucleotide charac-
ters (translating to 2,065,421 amino acid characters) from 4775
genes, represented by 12 ingroup species and two outgroup
species; human and mouse. Table 1 provides a list of all included
species and their sequence coverage in the alignment. After
eliminating potentially homoplastic sites, the alignment length was
reduced to 4,314,195 characters for the nucleotide data and
1,476,398 characters for the amino acid data. The average
sequences coverage of the alignment was 85.3%.
In heuristic analyses and RAxML parametric bootstrap analysis
the relationships within the orders were unanimous, but some
inter-ordinal relationships received only limited support. Figure 1
shows the best-supported tree and branch lengths based on
maximum likelihood (ML) analysis of concatenated amino acid
(AA) data. This topology was also the best or among the best
Figure 1. Best ML tree based on concatenated amino acid data.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org3December 2011 | Volume 6 | Issue 12 | e28199
supported in other analyses. For further evaluating the topology
and the support assigned to it, exhaustive analyses were performed
on the relationships among the five laurasiatherian orders, testing
all 105 possible rooted topologies. Among these 105 trees any
topology where Lipotyphla was not the first diverging order
received significantly lower support in all analyses. Thus, in the
following only the remaining 15 proposed trees among Chiroptera,
Cetartiodactyla, Perissodactyla, and Carnivora were analyzed in
The 15 topologies are shown and numbered in Figure 2. The
Shimodaira-Hasegawa probabilities (pSH)  for these 15
topologies are shown in Table 2. Tree 14 is favored by most
analyses and not rejected by any analysis. This tree corresponds to
that shown in Figure 1. While the AA and NT12 (nucleotides, first
and second codon position) datasets do not provide conclusive
support for a single tree, tree 14 is significantly supported by
NT123 (nucleotides, all codon positions). Removing the most
distant outgroup, opossum, from the analysis generally increases
the support for topology 14 relative to the others, illustrating the
importance of using a close outgroup for phylogenetic analyses.
The estimation of divergence times was complicated by the lack
of distinctive outgroups with a well-defined maximum age. Using
the soft lower bounds (Table S1) yielded unexpected ancient
divergence times among all groups. By constraining the deepest
divergence to 92 Ma  divergence times that are in agreement
with previous phylogenomic studies were estimated. Thus the
radiation among the different order occurred 87–60 Ma. While
the absolute dates may be debatable, the relative divergence times
of short branches that are problematic to resolve were in the order
of 2 Myr (Figure S1).
Topology 5 receives the second best support, joining Perisso-
dactyla and Cetartiodactyla , to the exclusion of Carnivora.
This hypothesis is the best supported in AA analyses, but clearly
rejected by NT123 data. There is no majority consensus among
the analyses or data sets. Interestingly, topology 8, termed
Pegasoferae  was favored in an earlier retroposon insertion
analysis, but receives low support by AA data using TF, and is
significantly rejected in all other sequence analyses. In general, the
lack of clear support for a single topology mirrors the results and
conclusions on the most basal placental mammalian divergences
. Therefore, network analysis methods were employed to
investigate the conflict in the data.
Figure 3 shows a consensus network based on 1159 trees
calculated from ML analyses of the alignments of single genes for
which sequence data for all 14 species were available. The
relationships between the four orders Carnivora, Chiroptera,
Cetartiodactyla, and Perissodactyla are largely unresolved in this
analysis, as represented by the cube-like structure in this part of the
network. For the lack of an acceptable name this clade will be
abbreviated by the initial letters of the orders as ‘‘CCCP-clade’’.
The cube-like structure illustrates the roughly equal support for
placing an order in either topology along parallel branches. In
other parts of the network, i.e. within Cetartiodactyla and
Carnivora, the relationships are depicted by elongated structures,
indicating that one of the topologies is preferred over the other by
The results of the individual gene trees were also analyzed, with
the summary shown in Table 3. The table shows, for each data set
(AA, NT12, NT123 and codon) and each of the 15 topologies, the
number of times the topology was among the ones with the
maximal likelihood value, how often it was rejected, and the sum
of the log-likelihoods. As in the analysis of the concatenated genes,
no consensus is found among methods and data sets, but a few
trends can be observed. Tree 5 always has the highest log-
likelihood sum and is rejected the fewest numbers of times in the
NT12 and NT123 data sets. It is also the most frequent ML tree in
the NT123 and codon data sets. For most of these combinations of
data and methods, tree 14 follows in second position. For certain
analyses on the AA and NT12 data set, trees 2, 10 and 12 have the
highest support, but all three of them are firmly rejected by other
analyses. It is also noteworthy, that the NT analyses, in particular
NT123, allow for a much stronger separations among the
Figure 2. Overview of the rooted topologies among four orders that have been individually tested.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org4December 2011 | Volume 6 | Issue 12 | e28199
Table 2. pSH values for 15 topologies regarding the relationship among Chiroptera, Perissodactyla, Carnivora, and Cetartiodactyla.
TF pSH(AA)TF pSH(NT12)TF pSH(NT123) RAxML (w/o opossom)
w opossum w/o opossumw opossum w/o opossumw opossum w/o opossumAANT12NT123
5 1.01.01.00.872550 0.0019 N/RBEST*
7 0.2779 0.391150000***
80.2663 0.3562 0.000950.000100***
10 0.135 0.26545 0.10780.013600 BEST**
120.066 0.001400 0.09110***
14 0.489950.587050.8414 1.0 1.01.0 N/R BESTBEST
Bold typeface indicate that the topology is not rejected. RAxML does not provide probability values and instead shows only if the topology is the most likely (BEST), not
rejected (N/R) or rejected at the 0.05 significance level (*).
Note – ‘‘w opossum’’ and ‘‘w/o opossum’’ denotes whether or not opossum was included as an outgroup. 0 denotes a probability below 0.0001.
Figure 3. Consensus network of 1159 trees based on alignments with sequence for all species, using a threshold value of 8%.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org5December 2011 | Volume 6 | Issue 12 | e28199
topologies. In the AA data set, the number of times a tree is among
the best ranges from 157 to 175, whereas for the NT123 data set, it
ranges from 86 to 181, a span that is more than five times as large.
Also, the highest log-likelihood difference for AA data is 233.0
compared to 977.5 for the NT123 data. This may be an indication
that the rate of amino acid substitution is often too low to
distinguish between the very short branches separating the orders,
while the numerous synonymous third codon positions may still
allow to better resolve some branches, despite their advanced state
of randomization at 80 Ma.
The analysis of the influence from sequence and alignment
properties on the resulting best topology is shown in Table 4. To
exclude irrelevant changes among less likely trees, only the five
most likely trees were compared. Some factors have an influence
on the results. Tree 10, for example, gets almost double the
support from long alignments than from short ones, whereas trees
12 and 14 find more support when longer distances separate the
sequences. Overall, the sequence distance has the largest effect on
the distribution of supported trees. Alignments with long distances
favor tree 14 and disfavor trees 2 and 5. Although this shows that
sequence specific aspects can influence the topology, none of the
chi-square tests indicate a significant difference between them.
The Bayesian species tree reconstruction, using the program
BEST, from 764 alignments selected for length and sequence
coverage did not yield a clearly supported bifurcating tree.
Posterior probabilities above 0.05 were calculated for three trees:
the probabilities were 0.43 for tree 15, 0.26 for tree 9 and 0.19 for
The alignments of the informative retroposon inserts are shown
in Figure S2. For the monophyly of the uncontroversial
laurasiatherian, Carnivora, and the cow-dolphin clade (Cetrumi-
nantia)  four to seven retroposon insertions were identified. In
addition, non-significant support, i.e. less than three retroposon
insertions  were found for the monophyly of the uncontro-
versial Cetartiodactyla, the CCCP-clade and the pig-cow-dolphin-
clade. The retroposon insertions that support uncontroversial
groupings are summarized in Table 5A and shown in Figure 4.
For these clades no contradictory signal from retroposon insertion
marker were identified.
Apart from these non-controversial markers, a number of
mutually incompatible retroposon insertions were found that
support different inter-ordinal relationships of the four orders
Table 5B summarizes the support from retroposon insertion data
Table 3. Analysis of 1159 gene trees.
1 154 423
3 159 418
2350.8 76 117
2429.9 90 122
5 1634020.02394180.0 120 880
2452.4 97 118
2135.2 99 92
2143.2 101 108
For each of the 3 data types (AA, NT12 and NT123) and tree topology, the number of times the topology was among the best (ML), the number of rejections and the
difference of the sum of the log-likelihoods to the best one are reported. Bold numbers indicate the best values in a column, while numbers in italics indicate the
respective second best values.
Table 4. Analysis of the influence of five aspects of the
alignments on the frequency of the five most likely trees.
25 1012 14
Alignment lengthshort 2225 14 20 252.86
long 23 262214 24
Longest distancelow26 271716 20 3.11
high 19241918 29
Sum of distances low19 2618 15 201.57
Codon usage biasaverage252518 21292.08
extreme 20 261813 20
Nucleotide usage biasaverage 24 2717 20241.19
extreme2124 19 1425
The numbers indicate how often each topology was the only highest supported
topology. With four degrees of freedom, none of the x2values are significant.
The bold number pairs indicate the largest change for each measure.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org 6December 2011 | Volume 6 | Issue 12 | e28199
for different topologies and Figure 5 depicts the network that can
be reconstructed from it. An equal number of three markers
support the hypotheses that Perissodactyla or Cetartiodactyla
represent the first divergence among the four orders, while two
marker support Carnivora as the first divergence. In addition two
markers support a grouping of Carnivora and Perissodactyla and
one marker group Carnivora with Chiroptera.
Compared to phylogenetic analyses that were done in the 1990s
and were based on single genes or a combination of a few
sequences, the advance in genome sequencing now make it
possible to analyze thousands of sequences, which promises a huge
increase in the accuracy of the reconstructed tree. In this study
4775 protein-coding sequences were used to reconstruct the
evolutionary history of a major clade of placental mammals, the
The best-supported tree in the phylogenomic analyses on
concatenated data among laurasiatherian orders (Figure 1)
conforms to that of previous mitogenomic and nuclear gene
analyses [11,14]. All analyses agree that Lipotyphla represent the
first divergence within this clade. Also, most sequence analyses can
significantly reject some hypotheses, such as the recently proposed
Pegasoferae hypothesis . However, the support for bifurcating
inter-ordinal relationships is surprisingly limited. Two incompat-
ible hypotheses of laurasiatherian relationships cannot be ruled out
and were even estimated to be the best ML tree in some analyses.
This indicates conflicting phylogenetic signals from the sequence
It has been suggested that separate analysis in which each gene
is evaluated individually is preferable to the analyses of
concatenated sequences, because this approach improves the
estimation of likelihood parameters [33,41]. Yet, separate analyses
of single genes lead to the same phylogenetic conclusions as the
analysis of concatenated data. The single gene analyses do not
favor a single tree but find support for alternative hypothesis, as
illustrated in the network of Figure 3. This point is nicely depicted
in the network of topologies reconstructed from single genes. Short
sequences, however, by their nature often do not contain enough
information to significantly distinguish between different topolo-
gies. A solution to this problem is the combination of likelihood
values from single gene analyses . This approach, like the
concatenated analyses, favors topologies 5 and 14 in all analyses.
In addition, the influence of key characteristics of the individual
sequences (such as sequence length, rate of evolution and
composition) on the reconstructed trees was investigated, because
different subsets of the data may have different reconstruction
biases, such as long branch attraction . Such biases would
cause substantial numbers of sequences supporting conflicting
topologies. Yet, none of the five tested characteristics had a
significant effect on the distribution of the favored topologies.
Although there are certainly additional, but untested properties of
the sequences or alignments, the outcome of this study supports
the idea that the data contain truly conflicting phylogenetic signals
rather than subsets of genes that are affected by different
reconstruction biases. The conflicting evolutionary signals from
single genes cannot be reconciled into a bifurcating tree, even
when using reconstruction methods that take coalescence models
into account. This method is supposed to allow reconstruction of a
Figure 4. Non-controversial retroposon markers shown on a
Table 5. Number (#) of retroposon markers supporting
relationships among Laurasiatheria.
a) Uncontroversial relationships#
Lipotyphla first divergence in Laurasiatheria2
b) Incompatible relationships#
((Carnivora+Perissodactyla), Chiroptera, Cetartiodactyla)
((Carnivora+Chiroptera), Perissodactyla, Cetartiodactyla)
Figure 5. Network of relationships supported by retroposon
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org 7 December 2011 | Volume 6 | Issue 12 | e28199
species tree despite the presence of incomplete lineage sorting, but
does not account for any sort of lateral gene transfer or
introgression through hybridization.
With current methods it is difficult to distinguish between
incomplete lineage sorting and introgression. However, both
processes have a profound effect on the definition of a species at
the genomic level, because it causes alleles to be shared between
species, which then contradict each other in delineating species or
estimate their divergences. Even over long time periods these
shared alleles are influencing phylogenetic reconstruction, despite
many new mutations, which are unique to each order. Thus, the
genomes of today’s orders, which started out 70 Ma as different
populations and then species, retain information of the past
Not only incomplete lineage sorting, but also hybridization
occurs more frequently in animals than previously assumed. While
hybridization has generally been considered to hinder evolution-
ary diversification , hybridization from distant populations or
other species can introduce novel mutation, increasing the
possibility for adaptation . Evidence for hybridization that
aid adaptation has been described in insects  and fishes ,
and is not unexpected to occur in birds and mammals, given the
frequency of hybridization in these . A number of hybridiza-
tion events in mammals have been described, indicating that it
may not be a rare process  and consequently hybridization in
animals gains an increasing interest .
Finally, the application of close outgroups has not only
increased the amount of data but also yielded more consistent
results compared to when a more distant outgroup, the opossum, is
used. This agrees with previous observations that suggest using a
closer outgroup often increases the level of support for the correct
The support for different topologies, as provided by individual
loci becomes obvious in the retroposon analysis. This study
focused on LINE 1 elements, which were active during this time of
placental mammalian evolution [16–18]. The conflicts in the
resolution of the relationships of the CCCP-clade by retroposon
data mirror the sequence-based analyses of these relationships. In
particular, divergences within the CCCP-clade for which the inter-
ordinal relationships were not clearly resolved by sequence data
analyses, were studied in detail by retroposon insertions. A number
of retroposon insertions for a possible Pegasoferae relationship
((Perissodactyla, Carnivora), Chiroptera) have been found, but
unlike in the study of Nishihara et al. , the current study
identified numerous conflicting retroposon insertion, supporting
alternative relationships (Figure 5). In comparison, well-resolved
relationships within Laurasiatheria are unambiguously resolved by
retroposon data (Figure 4). For these unambiguous groups no
contradictory signals were identified in this survey. The congruent
results from retroposon data and sequence based analyses, support
the view that the lack of resolution from sequence data is not
caused by systematic errors.
Retroposon insertion data are, with very few exceptions,
regarded as being homoplasy free [51–53]. The rarity and very
mechanism of retroposon insertion support the idea that retro-
poson insertion reversals or parallel events are non-existing or
extremely uncommon . However, these and previous findings
 show that retroposon insertion data can still produce
contradictory phylogenetic signals stemming from genomic events
that are connected with speciation, such as incomplete lineage
sorting or hybridization. In fact, apparently contradictory
sequence data and retroposon data in this study, along with that
provided by an investigation into the early placental mammalian
evolution may be best interpreted as a result of such processes .
However this leads to a problem when regarding the statistics of
branch support from retroposon insertions.
The premises for the hypothesis that three retroposon insertions
are sufficient to significantly support a branch  was that these
data are homoplasy free and do not produce conflicting data.
However, as outlined above evolutionary processes do produce
conflicting phylogenetic signals from retroposons, if one interprets
the data in a strictly bifurcating tree [2,5,6]. Thus, the simple
statistics that suggests that three retroposon insertions in one
branch yield significant support needs to be revised to include the
possibility of conflicting signal.
Sequence data and retroposon insertion data can lead to
apparently inconsistent hypotheses, when viewed as a bifurcating
tree. A sequence based tree analyses of concatenated sequences
represents only an average of the phylogenetic signal. Most
phylogenetic information can get lost or distorted. However, the
complex pattern from sequence and retroposon-based analyses
can better be depicted as networks  and easily explain
apparent inconsistencies and allow illustrating and exploring
conflicting data. This way apparent inconsistencies are naturally
resolved by making sense out of the complex evolutionary
patterns. By placing all events on separate branches, alternative
evolutionary pathways and gene-trees are revealed (Figure 5).
The problem of phylogenentic inconsistencies arises only when
one ignores the possibility of complex evolutionary history and
tries to force them into a traditional, two-dimensional bifurcating
tree. Complex evolutionary patterns or conflict of rare genomic
events have now been described for Laurasiatheria, hominoid
divergences , basal placental mammalian divergences [2,5],
and other mammalian lineages [55–57].
In all cases of complex speciations the divergence times among
the groups are relatively short . This is also the case for
Laurasiatheria in which the estimated times for some groups are
within about 2 Myr of each other. This is, as discussed earlier for
other divergences , within the order of speciation times of
divergence and species durations [58,59], which can lead to the
complex pattern of gene trees. Speciation related processes have
obviously influenced the evolution of placental mammals to a
much larger extent than expected and in many cases do not allow
the reconstruction of bifurcating divergences. Stochastic errors of
small datasets aside, conflicting trees that were in previous studies
based on small data sets may actually reflect alternative
evolutionary scenarios of single genes in the genome, resulting in
different gene trees . Although prokaryote evolution represent
an extreme case of network-like evolution [61,62], the evolution
and speciation of vertebrates may be more complex than
With the advent of more genome data becoming available,
along with the ability to explore deep divergences in greater detail,
it is becoming evident that evolutionary processes are best
interpreted as networks. Networks naturally highlight the conflict
and difficulties of previous phylogenetic studies to find a congruent
bifurcating tree within this group. The hope of phylogeneticists
that whole genome data would one day yield a single, stable and
bifurcating evolutionary tree [18,63,64], is not fulfilled for some
parts of the placental mammalian tree. However, it seems that a
more valuable lesson can be learned from genome analyses. That
is, some divergences are not characterized by bifurcations but
rather that the evolution of some placental mammals represent a
complex pattern of genealogies of different parts of the genome.
Speciation processes that can be revealed from genome data even
for deep divergences, define this pattern. The evolution of
Carnivora, Perissodactyla, Chiroptera, and Cetartiodactyla (Laur-
asiatheria) represent such a case.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org8December 2011 | Volume 6 | Issue 12 | e28199
Chronogram showing the estimated times of
insertions found in this study.
Alignments of all informative retroposon
points used in the divergence time estimation.
Upper and lower bounds of the calibration
We thank Matthew Hartfield for proofreading.
Conceived and designed the experiments: BMH AJ. Performed the
experiments: BMH. Analyzed the data: BMH AS SZ. Contributed
reagents/materials/analysis tools: BMH AS SZ. Wrote the paper: AJ
1. Hallstro ¨m BM, Janke A (2008) Resolution among major placental mammal
interordinal relationships with genome data imply that speciation influenced
their earliest radiations. BMC Evol Biol 8: 162.
2. Hallstro ¨m BM, Janke A (2010) Mammalian evolution may not be strictly
bifurcating. Mol Biol Evol 27: 2804–2816.
3. Maddison WP (1997) Gene trees in species trees. Syst Biol 46: 523–536.
4. Edwards S (2008) Is a new and general theory of molecular systematics
emerging? Evolution 63: 1–19.
5. Churakov G, Kriegs JO, Baertsch R, Zemann A, Brosius J, et al. (2009) Mosaic
retroposon insertion patterns in placental mammals. Genome Res 19: 868–875.
6. Nishihara H, Hasegawa M, Okada N (2006) Pegasoferae, an unexpected
mammalian clade revealed by tracking ancient retroposon insertions. Proc Natl
Acad Sci U S A 103: 9929–9934.
7. Waddell PJ, Okada N, Hasegawa M (1999) Towards resolving the interordinal
relationships of placental mammals. Syst Biol 48: 1–5.
8. Asher RJ, Helgen KM (2010) Nomenclature and placental mammal phylogeny.
BMC Evol Biol 10: 102.
9. Novacek MJ (1992) Mammalian phylogeny: shaking the tree. Nature 356:
10. Xu X, Janke A, Arnason U (1996) The complete mitochondrial DNA sequence
of the Greater Indian Rhinoceros, Rhinoceros unicornis, and the phylogenetic
relationship among Carnivora, Perissodactyla and Artiodactyla (+Cetacea). Mol
Biol Evol 13: 1167–1173.
11. Arnason U, Adegoke JA, Bodin K, Born EW, Esa YB, et al. (2002) Mammalian
mitogenomic relationships and the root of the eutherian tree. Proc Natl Acad
Sci U S A 99: 8151–8156.
12. Pumo DE, Finamore PS, Franek WR, Phillips CJ, Tarzami S, et al. (1998)
Complete mitochondrial genome of a neotropical fruit bat, Artibeus jamaicensis,
and a new hypothesis of the relationships of bats to other eutherian mammals.
J Mol Evol 47: 709–717.
13. Mouchaty SK, Gullberg A, Janke A, Arnason U (2000) The phylogenetic
position of the Talpidae within eutheria based on analysis of complete
mitochondrial sequences. Mol. Biol. Evol 17: 60–67.
14. Murphy WJ, Eizirik E, O’Brien SJ, Madsen O, Scally M, et al. (2001) Resolution
of the early placental mammal radiation using Bayesian phylogenetics. Science
15. Li R, Fan W, Tian G, Zhu H, He L, et al. (2009) The sequence and de novo
assembly of the giant panda genome. Nature 463: 311–317.
16. Kim TM, Hong SJ, Rhyu MG (2004) Periodic explosive expansion of human
retroelements associated with the evolution of the hominoid primate. J Korean
Med Sci 19: 177–185.
17. Kriegs JO, Churakov G, Kiefmann M, Jordan U, Brosius J, et al. (2006)
Retroposed Elements as Archives for the Evolutionary History of Placental
Mammals. PLoS Biol 4: e91.
18. Murphy WJ, Pringle TH, Crider TA, Springer MS, Miller W (2007) Using
genomic data to unravel the root of the placental mammal phylogeny. Genome
Res 17: 413–421.
19. Kramerov DA, Vassetzky NS (2005) Short retroposons in eukaryotic genomes.
Int Rev Cytol 247: 165–221.
20. Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy
and high throughput. Nucleic Acids Res 32: 1792–1797.
21. Dress AWM, Flamm C, Fritzsch G, Gru ¨newald S, Kruspe M, et al. (2008) Noisy:
Identification of problematic columns in multiple sequence alignments.
Algorithm Mol Biol 3: 7.
22. Jobb G, von Haeseler A, Strimmer K (2004) TREEFINDER: a powerful
graphical analysis environment for molecular phylogenetics. BMC Evol Biol 4:
23. Stamatakis A (2006) RAXML-VI-HPC: maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models. Bioinformatics 22:
24. Lanave C, Preparata G, Saccone C, Serio G (1984) A new method for
calculating evolutionary substitution rates. J Mol Evol 20: 86–93.
25. Whelan S, Goldman N (2001) A general empirical model of protein evolution
derived from multiple protein families using a maximum likelihood approach.
Mol Biol Evol 18: 691–699.
26. Benton MJ, Donoghue PCJ, Asher RJ (2009) Calibration and constraining
molecular clocks. In: Hedges SB, Kumar S, eds. The timetree of life. Oxford:
Oxford University Press. pp 35–86.
27. Drummond A, Rambaut A (2007) BEAST: Bayesian evolutionary analysis by
sampling trees. BMC Evol Biol 7: 214.
28. Suchard M, Rambaut A (2009) Many-core algorithms for statistical phyloge-
netics. Bioinformatics 25: 1370–1376.
29. Zoller S, Schneider A (2010) Empirical Analysis of the Most Relevant
Parameters of Codon Substitution Models. J Mol Evol 70: 605–612.
30. Clifford P (1994) In discussion of ‘‘Approximate Bayesian inference with the
weighted likelihood bootstrap’’ by Newton MA and Raferty AE. J Roy Statist
Soc B 56, 35.
31. Suchard M, Weiss R, Sinsheimer J (2001) Bayesian selection of continuous-time
Markov chain evolutionary models. Mol Biol Evol 18: 1001–1013.
32. Kass R, Raferty A (1995) Bayes factors. J American Stat Assoc 90: 773–795.
33. Pupko T, Huchon D, Cao Y, Okada N, Hasegawa M (2002) Combining
multiple data sets in a likelihood analysis: Which models are the best? Mol Biol
Evol 19: 2294–2307.
34. Huson DH, Bryant D (2006) Application of phylogenetic networks in
evolutionary studies. Mol Biol Evol 23: 254–267.
35. Liu L (2008) BEST: Bayesian estimation of species trees under the coalescent
model. Bioinformatics 24: 2542–2543.
36. Katoh K, Toh H (2008) Recent developments in the MAFFT multiple sequence
alignment program. Brief Bioinform 9: 286–298.
37. Shimodaira H, Hasegawa M (1999) Multiple comparisons of log-likelihoods with
applications to phylogenetic inference. Mol Biol Evol 16: 1114–1116.
38. Montgelard C, Catzeflis FM, Douzery E (1997) Phylogenetic relationships of
artiodactyls and cetaceans as deduced from the comparison of cytochrome b and
12S rRNA mitochondrial sequences. Mol Biol Evol 14: 550–559.
39. Arnason U, Gullberg A, Gretarsdottir S, Ursing B, Janke A (2000) The complete
mitochondrial genome of the sperm whale and the establishment of a new
molecular reference for estimating eutherian divergence dates. J Mol Evol 50:
40. Waddell PJ, Kishino H, Ota R (2001) A phylogenetic foundation for
comparative mammalian genomics. Genome Inform 12: 141–154.
41. Rannala B, Yang Z (2008) Phylogenetic inference using whole genomes. Annu
Rev Genomics Human Genet 9: 217–231.
42. Bergsten J (2005) A review of long-branch attraction. Cladistics 21: 163–193.
43. Mayr E (1963) Animal Species and Evolution. Cambridge: Belknap Press of
Harvard University Press.
44. Seehausen O (2004) Hybridization and adaptive radiation. Trends Ecol Evol 19:
45. Lewontin RC, Birch LC (1966) Hybridization as a source of variation for
adaptation to new environments. Evolution 20: 315–336.
46. Bell MA, Travis MP (2009) Hybridization, transgressive segregation, genetic
covariation, and adaptive radiation. Trends Ecol Evol 20: 358–361.
47. Mallet J (2007) Hybrid speciation. Nature 446: 279–283.
48. Gray AP (1972) Mammalian hybrids: a check-list with bibliography, 2nd ed
Commonwealth Agricultural Bureaux.
49. Schwenk K, Brede N, Streit B (2008) Introduction. Extent, processes and
evolutionary impact of interspecific hybridization in animals. Philos Trans R Soc
Lond B Biol Sci 363: 2805–2811.
50. Schneider A, Cannarozzi GM (2009) Support patterns from different outgroups
provide a strong phylogenetic signal. Mol Biol Evol 26: 1259–1272.
51. Steel M, Penny D (2000) Parsimony, likelihood, and the role of models in
molecular phylogenetics. Mol Biol Evol 17: 839–850.
52. Cantrell MA, Filanoski BJ, Ingermann AR, Olsson K, DiLuglio N (2001) An
Ancient Retrovirus-like Element Contains Hot Spots for SINE Insertion.
Genetics 158: 769–777.
53. van de Lagemaat LN, Gagnier L, Medstrand P, Mager DL (2005) Genomic
deletions and precise removal of transposable elements mediated by short
identical DNA segments in primates. Genome Res 15: 1243–1249.
54. Ray DA, Xing J, Salem AH, Batzer MA (2006) SINEs of a nearly perfect
character. Syst Biol 55: 928–935.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org9December 2011 | Volume 6 | Issue 12 | e28199
55. Ebersberger I, Galgoczy P, Taudien S, Taenzer S, Platzer M, et al. (2007)
Mapping human genetic ancestry. Mol Biol Evol 24: 2266–2276.
56. Janecka JE, Miller W, Pringle TH, Wiens F, Zitzmann A, et al. (2007) Molecular
and genomic data identify the closest living relative of primates. Science 318:
57. Kriegs JO, Churakov G, Jurka J, Brosius J, Schmitz J (2007) Evolutionary history
of 7SL RNA-derived SINEs in Supraprimates. Trends Genet 23: 158–161.
58. Curnoe D, Thorne A, Coate JA (2006) Timing and tempo of primate speciation.
J Evol Biol 19: 59–65.
59. van Dam JA, Abdul Aziz H, Alvarez Sierra MA, Hilgen FJ, Hoek Ostende LW,
et al. (2006) Long-period astronomical forcing of mammal turnover. Nature 443:
60. Nei M (1987) Molecular evolutionary genetics. New York: Columbia University
61. Doolittle WF (1999) Phylogenetic classification and the universal tree. Science
62. Kunin V, Goldovsky L, Darzentas N, Ouzounis CA (2005) The net of life:
Reconstructing the microbial phylogenetic network. Genome Res 15: 954–959.
63. O’Brien SJ, Menotti-Raymond M, Murphy WJ, Nash WG, Wienberg J, et al.
(1999) The promise of comparative genomics in mammals. Science 286:
64. Springer MS, Stanhope MJ, Madsen O, de Jong WW (2004) Molecules
consolidate the placental mammal tree. Trends Ecol Evol 19: 430–438.
Phylogenomics of Laurasiatherian Mammals
PLoS ONE | www.plosone.org10December 2011 | Volume 6 | Issue 12 | e28199