28 Building the Mammalian Sector of the Tree of Life Combining Different Data and a Discussion of Divergence Times for Placental Mammals

Back

Maureen A. O’Leary

Marc Allard

Michael J. Novacek

Jin Meng

John Gatesy

490

Mammals are species that comprise a clade formed by the

common ancestor of marsupials, monotremes, and placentals

and all of its living and extinct descendants. The oldest fossils

that form part of Mammalia (specifically the “crown clade”;

see below) are diminutive forms known as multituberculates

(Rougier et al.1996, McKenna and Bell 1997, Luo et al. 2002:

fig. 1) and members of a clade referred to as Australosphenida

(Luo et al. 2002; see also Flynn et al.1999, Rauhut et al.

2002), both of which date to the Middle Jurassic period.

Mammals inhabit all land masses of the world (Nowak 1999)

and have invaded such a wide range of habitats that they

currently can be found living in the air and the sea, on land

and within it. To exploit these habitats different mammalian

taxa have evolved into the largest animals ever to have inhabited

the earth (the blue whale), some of the most intelligent

forms of life based on the ratio of brain size to body size

(e.g., humans and chimpanzees), and forms possessing such

extraordinary behaviors as the ability to echolocate (e.g.,

certain bats) as a means of understanding their surroundings

(Nowak 1999). Building the mammal part of the Tree

of Life amounts to discovering the branching diagram (phylogenetic

tree) that describes how fossil and living mammal

species diversified from a common ancestor through time.

The living members of Mammalia possess a variety of

anatomical characteristics, including mammary glands, a

specialized skin gland that can produce milk to feed offspring.

Most living mammals, and some extraordinarily wellpreserved

fossil mammals (e.g., Hu et al.1997, Meng and

Wyss 1997, Ji et al. 2002) also have hair. Hair serves multiple

functions in mammals, including insulation, camouflage,

and display. The circulatory system of mammals is

characterized by a four-chambered heart consisting of fully

separated venous and arterial circulation, and the mammalian

brain includes dramatically expanded areas of gray

matter (Vaughan 1986), as well as highly developed centers

for processing visual and olfactory stimulation. Bony features,

such as the presence of three ear ossicles and a jaw joint,

known as the dentary-squamosal joint (Olson 1944, 1959,

Kermack and Mussett 1958, Simpson 1960, Crompton and

Jenkins 1973, 1979, Kermack et al.1973, Hopson 1994,

Crompton 1995, Cifelli 2001), also have served to diagnose

mammals, particularly fossil mammals. Different subgroups

of mammals also are famous for their diversity of dental specializations,

such as the tribosphenic molar (see below).

Researchers investigating mammalian phylogenetics

have pursued a range of questions from such focused tasks

as understanding the relationships of several closely related

species, to broad investigations of interordinal relationships,

many of which have involved the interpretation of numerous

key fossils. As is discussed throughout this volume, as

part of the Tree of Life effort, simultaneous analyses of hundreds

or even thousands of taxa are emerging, often referred

to as supermatrix analyses. The last decade has been characterized

by an enormous increase in the amount of moBuilding

the Mammalian Sector of the Tree of Life 491

lecular sequence data available for the study of mammal

phylogenetics as well as the discovery of a number of very

significant new fossils. As we discuss below, mammalian

phylogenetics is now moving away from a pattern of investigating

how and why there is incongruence between data

partitions (e.g., “molecules vs. morphology”) and toward

large-scale integration of historically heterogeneous character

data (e.g., osteology, histology, molecular sequences,

behavior). This approach often facilitates the discovery of

clades supported by characters that may come from many

different aspects of the organism. The clade Mammalia is

poised to become one of the first Linnaean classes to be examined

using a global simultaneous analysis of molecular

and phenotypic data because Mammalia is a clade of relatively

low taxonomic diversity relative to examples like Insecta

or Aves, and because it includes many species,

including our own (Homo sapiens), which are particularly

well characterized from both a molecular and a morphological

standpoint.

In this chapter, we do not provide a historical or taxonomic

review of work on mammal phylogenetics—this has

been provided recently for fossil data (Cifelli 2001) and for

molecular data (Waddell et al. 1999, and references therein).

Instead, we describe current efforts to move toward simultaneous

analysis of phylogenetic data for mammals as an

exemplar clade that forms part of the Tree of Life. We discuss

some of the methodological justification for simultaneous

analysis and explore particular problems within

mammalian phylogenetics, primarily focusing on questions

of interordinal relationships, because these have historically

been some of the most challenging and contentious problems.

Finally, as an example of how phylogenies can be applied

to other evolutionary questions, we discuss using

phylogenies to determine the age of placental mammals.

How Many Mammals Are There?

Zoologists now have recognized more than 5000 extant species

of mammals (Wilson and Reeder in press), but no contemporary

tally of the number of extinct mammal species has

been conducted. A count of genera, extinct and extant, can

be obtained from the recent classification of McKenna and

Bell (1997) and was reported by Shoshani and McKenna

(1998) to be 1083 living genera and 4076 extinct genera. Not

only are the majority of mammalian genera extinct, but for

every one extant genus of mammals, there are almost four

extinct genera (fig. 28.1). We estimate that a count of species

that included both extinct and extant taxa might uncover,

conservatively, 20,000 species. With so much extinction

recorded by fossils, this diversity must be accounted for in

building a phylogenetic tree for mammals. Put another way,

a tree based on living mammals alone encompasses only a

fraction of the known diversity of Mammalia.

Mammal Clades and Broad-Level Classification

Figure 28.2 illustrates the tripartite division of Mammalia into

Monotremata (the echidnas and duck-billed platypuses; species

that lay eggs rather than produce live young), Marsupialia

(kangaroos, opossums, koalas, and relatives), and Placentalia

(elephants, whales, primates, shrews, mice, dogs, bats, and

relatives). We have organized this tree using crown clade and

stem clade concepts (Jefferies 1979, Ax 1987, Rowe 1988,

de Queiroz and Gauthier 1992, Wible et al.1995, Rougier

et al.1996, McKenna and Bell 1997) as a means of defining

Mammalia and the clades within it. The use of crown and

stem clades as a final basis for classification remains controversial

for a variety of reasons (e.g., McKenna and Bell 1997,

Nixon and Carpenter 2000). Nonetheless, as we discuss

below, certain issues in mammalian phylogenetics, such as

the timing of the origin of Placentalia, have been muddled

by the inconsistent use of clade names by different authors.

Different authors often mean different species when they use

the word “placental,” particularly when referring to fossil taxa

and their relationships to clades of living taxa. We use crown

and stem clade concepts here because, lacking another unambiguous,

widely agreed upon (or used) mammalian classification,

these terms serve as an effective heuristic device

for discussing recent problems in mammalian phylogenetics,

such as the time of origin of various clades.

We can define the crown clade Placentalia as the common

ancestor of Elephas maximus (an elephant), Bos taurus

(domestic cow), and Dasypus novemcinctus (an armadillo),

and all of its living and fossil descendants. Some of the descendants

that form part of Placentalia include humans and

the rest of the order Primates, as well as a number of other

orders (fig. 28.1) containing such taxa as bats, anteaters, flying

lemurs, whales, carnivores, elephants, hippos, and tree

shrews, to name a few examples. Although crown clades are

defined by, and include, many extant species, it is important

to consider that crown clades also contain fossil species. For

example, the entirely extinct clade Desmostylia is part of the

crown clade Placentalia because desmostylians have been

demonstrated to be closely related to such species as Elephas

maximus and relatives (Domning et al.1986, Novacek and

Wyss 1987, Novacek 1992a) and therefore constitute descendants

of the common ancestor of Elephas maximus, Bos taurus,

and Dasypus novemcinctus. Many crown clades include numerous

fossils; in fact, fossils may be the majority of species

in crown clades, as is the case for placental mammals

(fig. 28.1).

The crown clade Marsupialia is defined as the common

ancestor of Didelphis virginiana (an opossum) and Macropus

giganteus (a kangaroo), and all of its living and fossil descendants.

The crown clade Monotremata is defined as the common

ancestor of Ornithorynchus anatinus (a platypus) and

Tachyglossus aculeatus (an echidna), and all of its living and

fossil descendants. These crown clades contribute to the

Figure 28.1. A summary of generic-level extinction within Mammalia. Taxonomic categories listed within Mammalia are primarily,

although not exclusively, the rank of Order (from McKenna and Bell 1997). We have summarized the numbers of extinct and extant

genera contained within each category with a reconstruction of an example species on the left. A skeleton or a jaw represents

categories with no living members; a silhouette of an example living species represents categories that have at least one living

member. In general, jaws indicate relatively less documented taxa; however, many species in the categories represented here by jaws

are also known from skulls and postcranial data. Note that the numbers of extinct genera far exceed the numbers of extant genera and

that many categories with extant members contain a majority of fossil taxa. Higher groupings (Monotremata, Placentalia, Theria,

Metatheria, Marsupialia, and Eutheria) follow figure 28.2 and are not necessarily those of McKenna and Bell (1997). For example,

McKenna and Bell’s (1997: 80–81) Placentalia of indeterminate order and Epitheria of indeterminate order are listed here as

“Eutheria, order indet.” McKenna and Bell (1997) classified some Cretaceous taxa within groups marked here as part of Placentalia

492

(e.g., Ungulata order indet., Meridiungulata). Contra Springer et al. (2003), however, this is not tantamount to evidence that the

crown clade Placentalia (fig. 28.2) contains Cretaceous taxa because the McKenna and Bell (1997) classification is not based on a

phylogenetic analysis. It should not be assumed, therefore, that groupings described in McKenna and Bell (1997) are necessarily

monophyletic because in many cases this remains to be explicitly tested. Taxa listed as Eutheria and Metatheria are stem clades to

Placentalia and Marsupialia, respectively (fig. 28.2); the category “Stem taxa to Theria” has not been formally named. Figures redrawn

from Sinclair (1906), Scott (1910), Riggs (1935), Simpson (1967), Clemens (1968), Kermack et al. (1968), Krebs (1971), Casiliano

and Clemens (1979), Kielan-Jaworowska (1979), Jenkins and Krause (1983), Dashzeveg and Kielan-Jaworowska (1984), Rose

(1987), Fox et al. (1992), Rougier (1993), Kielan-Jaworowska and Gambaryan (1994), Cifelli and DeMuizon (1997), Hu et al.

(1997), Novacek et al. (1997), Cifelli (1999), Ji et al. (1999), Pascual et al. (1999), and DeMuizon and Cifelli (2000).

493

494 The Relationships of Animals: Deuterostomes

definition of two larger crown clades: Theria, which is the

common ancestor of Elephas maximus and Didelphis virginiana

and all of its descendants; and Mammalia itself, which is the

common ancestor of Elephas maximus and Ornithorynchus

anatinus and all of its descendants. Theria is also a crown clade

for which the majority of taxa are fossils.

The terms Prototheria, Metatheria, Eutheria, and Mammaliaformes

represent stem-based taxa (De Queiroz and

Gauthier 1992). They are defined as follows: Metatheria consists

of all species more closely related to the marsupial Didelphis

virginiana than to the placental Elephas maximus; Eutheria

consists of all species more closely related to the placental

Elephas maximus than to the marsupial Didelphis virginiana;

Prototheria consists of all species more closely related to

Ornithorynchus anatinus than to Elephas maximus (in other

words, to any member of Theria); and Mammaliaformes consists

of all taxa more closely related to Elephas maximus than

to the clade Reptilia (sensu Gauthier et al. 1988).

It is important to recognize that this method of defining

larger clades implies that all placentals are also eutherians,

and alternatively that it is possible to be a eutherian without

being a placental (fig. 28.2). Any stem species that falls outside

the crown clade Placentalia would not be considered a

“placental” mammal; no matter how “placental-like” it is in

terms of its characters. The same would apply to other stem

species throughout the mammalian family tree. The recognition

of both crown and stem taxa depends on the pattern

of ancestry and descent, not on the characters that diagnose

a particular taxon. It may sound counterintuitive to define a

clade based on something other than anatomical characters;

however, as discussed by Rowe (1988), defining a clade by

common ancestry is consistent with evolutionary thinking

and allows so-called defining traits to reverse without disqualifying

a species’ membership in a larger clade. For example,

if we defined species as mammalian because they have

extensive hair, then strictly speaking, we would be barred

from considering whales (which virtually lack hair) as part

of Mammalia. This, however, contradicts findings from phylogenetic

analyses that indicate that whales are deeply nested

within placental mammals. Likewise, there is evidence that

extinct pterosaurs from the Mesozoic had hair (Padian and

Rayner 1993, Unwin and Bakhurina 1994). The membership

of pterosaurs within Mammalia has never emerged from any

phylogenetic analysis because pterosaurs share a greater

number of traits with the clade Archosauria (crocodilians,

dinosaurs, birds, and relatives) than they do with Mamma-

Figure 28.2. Simplified

schematic of the tripartite

division of mammals indicating

crown clades (Placentalia,

Marsupialia, Monotremata,

Theria, and Mammalia) and stem

taxa to these crown clades. The

horizontal gray line indicates

the Recent. Prototherians,

eutherians, metatherians, and

mammaliaforms are stembased

taxa (e.g., De Queiroz

and Gauthier 1992). Different

branch lengths are a reminder

that the fossils are staggered in

time throughout these clades.

Lineages that are fully extinct are

denoted by a dagger. The stem

clade to the crown clade Theria

that is indicated by an asterisk

is currently unnamed. Note that

these taxa would not be

considered “therians.”

prototherians

but not

monotremes

mammaliaforms

but not

mammalians

Prototheria Metatheria Eutheria

Theria

Mammalia

Mammaliaformes

Extant

outgroup Monotremata Marsupialia

metatherians

but not

marsupials eutherians

but not

placentals

Placentalia

*

Building the Mammalian Sector of the Tree of Life 495

lia. As noted in McKenna and Bell (1997), this does not mean

that crown and stem clades do not have diagnostic characters;

they do. Because characters show homoplasy, however,

the topology, not the character, is used to define the group.

Deciphering the interrelationships of placental orders (how

placental species form larger clades) has remained a challenging

problem for morphological systematists despite decades

of study (e.g., Novacek 1992b, Shoshani and McKenna 1998).

In the early 1990s several hypotheses of interordinal relationships,

derived initially from anatomical data, were beginning

to be intensively tested with new molecular data. The taxon

within Placentalia that branched first was generally considered

to be an edentate, and the following clades were thought to

be monophyletic (although alternative arrangements certainly

remained under consideration): Glires (rodents + rabbits);

Paenungulata (hyraxes + elephants + sea cows); Archonta (primates,

bats, tree shrews, and flying lemurs), and Ungulata

(hoofed mammals) (discussed in Novacek 1992b). Nonetheless,

the base of Placentalia remained rather bushlike

with certain nodes appearing repeatedly but with other

higher groupings remaining unstable and often supported

by only a few synapomorphies.

The infusion of numerous molecular sequence characters

into mammalian phylogenetics, particularly during the last

decade, introduced a variety of new phylogenetic hypotheses.

Controversies developed concerning the monophyly of several

higher clades, including insectivorans (moles, shrews, and

relatives), archontans, ungulates, glires, rodents (mice, voles,

and close relatives), and artiodactylans (hoofed mammals with

an even number of toes). Indeed, molecular data have even

resulted in trees that did not support the fundamental tripartite

division of mammals, instead associating monotremes and

marsupials as sister taxa to the exclusion of placentals (Janke

et al.1994, 1997, 2002; see also arguments in the morphological

literature: Gregory 1947, Kьhne 1973). New higher clades,

notably Afrotheria (golden moles, elephant shrews, elephants,

aardvarks, hyraxes, tenrecs, and manatees) and Laurasiatheria

(whales, artiodactylans, carnivorans, perissodactylans, pangolins,

bats, and several insectivorans) have been first proposed

on the basis of entirely molecule-based analyses (e.g., Murphy

et al. 2001, 2002, Madsen et al. 2001). These clades represent

a fairly fundamental restructuring of the placental branching

sequence that had not been previously supposed from an investigation

of morphological data. New molecule-based hypotheses

remind systematists that even long-held notions of

relationship are hypotheses that can be overturned by new

data. In many cases, however, the combined analyses of molecular

and morphological data required to test these new

hypotheses are only just emerging.

The Importance of Combining Data

Many ideas about the evolution of mammals have at their core

an implicit or explicit hypothesis of genealogical history. Such

commonly discussed topics as scenarios of adaptation or

notions of the age or place of origin of a taxon are fundamentally

dependent on a hypothesis that states how mammal

species branched from each other through time. A weakly

tested hypothesis of relationship is a shaky scaffolding for

all other inferences placed on top of it, underscoring the

importance of the Tree of Life to evolutionary study as a

whole.

Phylogenetics, like many historical sciences, stands in

contrast to experimental sciences because the hypotheses to

be tested (e.g., the origin of primates) are not experiments

that can be repeated (like the function of a particular enzyme

in a human cell). Students of historical problems can, however,

establish tests of a different kind: they can formulate a

hypothesis of relationships about how species branched from

each other through time, and test that hypothesis by looking

for evidence to reject it. Phylogenetic hypotheses (e.g.,

trees) start as nothing more than a guess about the roadmap

of evolution; these guesses are then tested against the biological

and paleontological evidence available. They are rejected

if they do not efficiently explain all of the data.

How do we know when we have the tree of mammals?

Essentially, like any hypothesis, we can always continue to

test it. When a hypothesis of relationship has been tested and

retested by adding new data (e.g., molecular sequences,

bones, soft tissues, behavior) to a global parsimony analysis,

and the hypothesis remains unchanged, the hypothesis is the

best explanation of all the data. We could then refer to such

a hypothesis as robust or stable (Nixon and Carpenter 1996).

It is important to appreciate that such well-tested trees may

not necessarily have high support measures, such as bootstrap

values or Bremer support. Once we establish a phylogenetic

tree that is stable to the addition of new data (i.e., we

add new data and the tree does not change), we can examine

how the characters used to build the tree changed through

time (using a method called optimization), as well as what

the tree says about the age and place of origin of various

clades.

A phylogenetic tree of mammals is tested by studying

heritable, phylogenetically independent traits (molecular,

morphological, or behavioral), called characters. Arguments

that certain characters are “misleading” and should be eliminated

before a tree is even tested are circular because if we

do not know the tree (which is built from characters) then

we do not know which characters are “good” and which are

“bad” before we test them. Some character systems that have

been described as “bad” include mitochondrial DNA

(mtDNA; e.g., Naylor and Brown 1998, Luckett and Hong

1998), teeth (e.g., Naylor and Adams 2001, Cifelli 2000), and

all morphological evidence (e.g., Hedges and Maxson 1996).

Others that have been described as “good” include cranial

characters (e.g., McDowell 1958), certain nuclear genes, and

molecular markers known as SINES (e.g., Nikaido et al.1999,

2001). We would argue that empirical work is currently not

extensive enough to substantiate such generalizations, some496

The Relationships of Animals: Deuterostomes

thing that is underscored by there being a number of exceptions

to all the examples listed here as being “good” or “bad.”

But what if we have tested some characters; can we generalize

on the basis of those results? If a character has been found

to have a lot of homoplasy in worms, should we assume that

it also has a lot of homoplasy in mammals? Because there are

examples where such generalizations have failed, and because

characters can be informative even if they show homoplasy,

it seems premature to introduce such assumptions into phylogenetic

analyses.

Also problematic for the rigorous testing of phylogenies

is a tendency to assert a priori that a particular set of characters

is correlated functionally or developmentally, and that

these characters will promote a misleading phylogenetic signal

if they are included in an analysis. Often this is enforced

by constraining tree topology, and using that tree to evaluate

the informativeness of new data. This is problematic for

at least two reasons: (1) functional correlation and phylogenetic

correlation are not necessarily the same thing, and one

should therefore not be used as a basis for assuming the other

(Farris 1969), and (2) a priori lumping of characters into

functional complexes forces them to group together in phylogenetic

analysis; breaking them into individual characters

does not. The latter allows the data themselves to indicate

whether the individual parts of the proposed “complex” even

evolved at the same node.

Finally, we might ask which species are necessary to

sample in order to find the tree of mammals. If our goal is

to discover the Tree of Life, then ultimately we want to know

where all species, fossil and living, fit on the tree. This does

not mean that all analyses must include all species from the

outset, but it does mean that we should think about the

problem of mammalian phylogenetics on a large scale and

cumulatively, with the results of each analysis open to testing

by adding new species and characters. Historically, many

of the earliest cladistic analyses of mammals contained relatively

few taxa by contemporary standards and used higher

taxa as operational taxonomic units (OTUs; e.g., Novacek

1982, 1992a, Rowe 1988). As computerized search algorithms

have become increasingly powerful, more recent

analyses have tended to use genera (e.g., Rougier et al.1998,

Murphy et al. 2001a, Gatesy and O’Leary 2001, Meng et al.

2003). Malia et al. (2003) emphasize, for trees to be maximally

accountable to the data, there ultimately should be a

final shift toward sampling at the species level as analyses

become increasingly exhaustive and algorithms become

more powerful.

Supertrees and Supermatrices

Separating characters into groups such as different genes,

genes and morphology, or types of morphology is known as

data partitioning (Kluge 1989, Nixon and Carpenter 1996).

This approach is sometimes explicitly preferred for phylogeny

reconstruction, and trees from separate analyses may

subsequently be combined into a summary tree. This approach

has been formalized as “supertree” methods (Sanderson

et al.1998), and supertrees have been constructed to investigate

higher clades of mammals (e.g., Liu et al. 2001). Alternatively,

data partitions can simply be combined in a single

simultaneous analysis, or a “supermatrix” (e.g., Murphy et al.

2001, Gatesy et al. 2002, Malia et al. 2003), the analysis of

which relies on character congruence or agreement among

individual characters (Kluge 1989, Nixon and Carpenter

1996). Partitioning data matrices, determining the separate

tree results, and then summarizing the shared topological

patterns using consensus techniques was a sequence of

operations originally called taxonomic congruence (Kluge

1989). The supertree approach is not the same as the taxonomic

congruence approach in which traditional consensus

techniques (e.g., Adams consensus, strict consensus,

majority rules consensus) are used to summarize results.

Rather than comparing shared clusters of taxa directly, commonly

used supertree methods, such as matrix representation

with parsimony (Baum 1992, Ragan 1992), recode

separate phylogenetic trees into a new data matrix that represents

these trees. Then the supertree matrix is analyzed

using parsimony algorithms. Various implementations of

matrix representation with parsimony differ in how they

move from original tree topologies to a coded data matrix,

how “characters” are weighted, and how conflicts are reconciled

(Baum 1992, Ragan 1992, Baum and Ragan 1993, Purvis

1995, Ronquist 1996, Sanderson et al.1998). Salamin et al.’s

(2002) recent supermatrix of grasses provides a review of

these methods.

Kluge (1989) criticized the separate analysis of data partitions

on numerous grounds (see also Nixon and Carpenter

1996, DeSalle and Brower 1997). Others support separate

analyses of data partitions, and there have been several criticisms

and rejoinders on this topic (Bull et al.1993, Miyamoto

and Fitch 1995, Nixon and Carpenter 1996, Cunningham

1997). These arguments are primarily directed at separate data

analysis, not supertrees specifically, but the general arguments

are relevant to evaluating the usefulness of supertrees.

Perhaps the primary inadequacy of the supertree approach

is its insensitivity to hidden support (Barrett

et al.1991) among the characters of different matrices used

in separate analyses (see Wilkinson et al. 2001, Pisani and

Wilkinson 2002). The simplest way to avoid the problematic

issues of weighting and redundancy in supertree analysis

is to include each character state once in a supermatrix

and let character congruence determine the best-supported

tree topology (Farris 1983, Kluge 1989). There is no theoretical

difference between the analysis of one mammal clade,

or the analysis of all mammals, or the analysis of the Tree of

Life. The only difference is one of scale. Therefore as we begin

to solve computational problems that have limited the

scale of phylogenetic analyses (e.g., the number of taxa that

can be analyzed), the construction of supermatrices rather

than supertrees becomes increasingly compelling.

Building the Mammalian Sector of the Tree of Life 497

Supermatrices of Extinct and Extant Taxa: Computational

Issues and Missing Data

It is well known that as the number of taxa increases, so does

the difficulty of the phylogenetic problem. For four taxa there

are three unrooted networks possible; for 14 taxa there are

more than 316 billion possible unrooted networks, and this

number rapidly increases (Kitching et al.1998: 41). Under the

existing paradigm of finding the most parsimonious solution

for each problem, large matrices require an extremely large

amount of computational power. Building a simultaneously

analyzed tree for all living and extinct mammals, conservatively

20,000 species, represented by tens of thousands of characters,

will pose substantial computational challenges.

Many large matrices have been examined using PAUP*

(Swofford 2000); however, several investigators (Nixon 1999,

Goloboff et al.1999, Goloboff and Farris 2001) have commented

that PAUP* exhibits severe limitations regarding the

rapidity of parsimony search strategies if a data matrix contains

more than 400 taxa. Computational challenges such as

these represent an active area of research (e.g., DeSalle et al.

2002, Janies and Wheeler 2002), and new search algorithms

such as POY (Wheeler and Gladstein 2000), the parsimony

ratchet (Nixon 1999) implemented through WinClada

(Nixon 2002) and NONA (Goloboff 1994), and TNT

(Goloboff, et al.1999, Goloboff and Farris 2001) have allowed

investigators to execute some of the largest parsimony-based

searches with relative efficiency. Both WinClada/NONA and

TNT software, for example, regularly produced large trees from

matrices containing more than 1700 OTUs (Allard et al. 2002).

These new methods represent a considerable advance for the

examination of large data sets.

Results of some of the largest phylogenetic analyses published

to date were determined using these new rapid heuristic

methods but still include fewer than 1000 taxa;

examples include eukaryotes (440 taxa; Lipscomb et al.1998)

and seed plants (500 taxa; Rice et al.1997, Nixon 1999, Janies

and Wheeler 2001). An intraspecific analysis of humans that

included 1771 individuals (Allard et al. 2002) was tested

in a phylogenetic framework using WinClada/NONA. For

mammals two examples of particularly large published matrices

included 264 taxa (for molecular data using the program

POY; Janies and Wheeler 2001) and 91 taxa (molecular

and morphological data, using PAUP*; Gatesy et al. 2002).

A supermatrix of mammals, particularly one that combines

molecular and nonmolecular (morphology, behavior)

characters, will have substantial missing data. Missing data

may occur because no investigator has scored a particular set

of taxa and characters for a clade, because a feature has

changed so much as to be absent, because a character is inapplicable,

or because a feature has not been preserved for

study (in a fossil, e.g., see discussion in Gatesy and O’Leary

2001, Kearney 2002). Operationally, some investigators create

composite taxa (e.g., a combination of two or more species

to make a genus level OTU) expressly to reduce missing

data. A composite behaves in a tree search as a single taxon,

the assumption being that it is monophyletic with respect to

the other taxa in the analysis. Composite taxa (and the implicit

monophyly assumptions they encode) can, however,

have serious effects on the resulting topology. A recent review

(Malia et al. 2003) of a mammalian supermatrix that

included composite taxa (Madsen et al. 2001) showed that

the construction of composite taxa did have dramatic and

not necessarily beneficial effects on tree topology. Prendini

(2001) and Malia et al. (2003) recommend breaking all composites

above the species level, an approach consistent with

recent arguments that missing data do not create false or

misleading evidence (e.g., Kearney 2002).

Specific Problems in Mammal Phylogeny

Mammaliaformes

There are a number of interesting fossil taxa that fall outside

the crown clade Mammalia but that are part of the clade

Mammaliaformes (fig. 28.2). These fossils capture critical

stages in the transition from an amniote sister taxon of mammals

to the crown clade Mammalia and include such groups

as Sinoconodontidae, Morganucodonta, Docodonta, and

Haramiyoidea (McKenna and Bell 1997). These close mammal

relatives are of generally small size and are known from

fossils that date back to the Triassic and Jurassic. Support

for the hypothesis that these taxa belong outside of crown

clade Mammalia has been demonstrated in many different

analyses, (Rowe 1988, Wible et al.1995, Rougier et al.1996,

Hu et al.1997, Ji et al.1999, 2002, Luo et al. 2001a, 2001b,

2002, Wang et al. 2001, Rauhut et al. 2002).

As described above, Mammaliaformes contain not only

these basal forms, but also the crown clades of monotremes,

marsupials and placentals. Investigation of general mammaliaform

relationships also concerns the diversification of

fossils that are more highly nested. Some of the most researched

mammaliaform problems include the position of:

(1) multituberculates (mentioned above as critical for dating

the entire mammal crown clade), (2) “triconodonts” and

their relation to monotremes and therians, and (3) the relationship

of monotremes to other Mesozoic mammals.

Parsimony analyses have commonly placed multituberculates

within the crown clade Mammalia (Rowe 1988, Wible

et al.1995, Rougier et al.1996, Hu et al. 1997, Ji et al. 1999,

Luo et al. 2001b, 2002, Wang et al. 2001). Multituberculates

may be either stem monotremes (Luo et al. 2001b, Wang et al.

2001), more closely related to therians (Luo et al. 2002, Rauhut

et al. 2002), or part of an unresolved polytomy with therians

and monotremes (Wible et al.1995). Phylogenetic investigations

of “triconodonts” [Austrotriconodontidae, Amphilestidae,

and Triconodontidae (McKenna and Bell 1997)], a group that

may not be monophyletic, have shown them to be the sister

group of the crown clade Mammalia (Hu et al.1997, Luo et al.

498 The Relationships of Animals: Deuterostomes

2001b, 2002, Ji et al. 2002) or a member of the crown clade

Mammalia (Rowe 1988, Wible et al.1995, Rougier et al.1996,

Luo et al. 2001a 2002) or have been unable to resolve their

position with respect to crown Mammalia (Luo et al. 2001b,

Wang et al. 2001). When “triconodonts” fall within Mammalia,

they commonly form the sister group of the clade consisting

of multituberculates and their relatives (Rowe 1988, Luo

et al. 2002: fig. 1, Ji et al. 2002, Rauhut et al. 2002). It is clear

that phylogenetic relationships of triconodonts and multituberculates

are still unstable and that the resolution of this

problem will affect the content and relationships of Mammalia

as well as minimum estimates of the age of this clade.

Understanding the relationships of monotremes to other

mammals requires discussion of the term tribospheny.

Tribospheny refers to a shape of the molar teeth such that in

occlusion there is both a crushing (sphene) and a shearing

(tribos) component to the movement between the upper and

lower teeth (Simpson 1936). This dental feature is present to

varying degrees in the monophyletic group Theria (fig. 28.2)

and their close relatives; the oldest fossils of these have generally

been found in the Northern Hemisphere. Recent discoveries

of Mesozoic mammals from southern continents (Rich

et al.1997, Flynn et al.1999), however, have raised intriguing

questions about the origin of tribospheny and the relationships

of Mammalia. Several fossil taxa from southern continents

show a complex molar pattern that is tribosphenic in shape.

Rich et al. (1997, 1999, 2001a, 2001b) interpreted Ausktribosphenos

and Bishops, fossil mammals from the Cretaceous

period of Australia, as having tribosphenic molars and suggested

that these fossils were closely related to hedgehoglike

placentals. This affiliation for Ausktribosphenos and Bishops,

however, has not yet been corroborated by comprehensive

cladistic analyses (Kielan-Jaworowska et al.1998, Musser and

Archer 1998, Archer et al.1999, Rich et al. 2001a, 2001b).

Archer et al. (1999) interpreted the dentition of the Early Cretaceous

taxon Steropodon, a fossil monotreme, as having a

modified tribosphenic pattern. Monotremes are toothless in

the living adult forms, prohibiting direct comparison of their

adult dentition teeth with that of other mammals. The significance

of theses observations is that depending on the phylogenetic

hypothesis, monotremes may be descended from

a taxon with tribospheny, a character that has long been

thought to be more typical of therians. Recent phylogenetic

analyses also suggest that the tribosphenic molar pattern,

long thought to have evolved once, has evolved independently

twice: once in an endemic southern (Gondwanan)

clade that is survived by extant monotremes and again independently

in a northern (Laurasian) clade composed of extant

marsupials, placentals, and their extinct relatives (Luo

et al. 2001a, 2002, Ji et al. 2002, Rauhut et al. 2002).

Glires

Within the diverse clades that form part of placental mammals,

certain clades have been the focus of numerous investigations

drawing on morphological, molecular, and fossil

evidence. One such clade is Glires, which consists of two

extant mammalian orders: Rodentia (rats, mice, and relatives)

and Lagomorpha (hares and pikas). We define these here as

crown clades (fig. 28.3). Together, lagomorphs and rodents

constitute nearly half of extant mammalian species diversity

(Nowak 1999). Stem taxa to Lagomorpha first appeared in

the Paleocene of Asia (McKenna 1982; fig. 28.3). Crown

clade Rodentia, by contrast, may contain some taxa collected

in Paleocene rocks of North America (Wood 1962, Dawson

et al.1984, Korth 1984, Dawson and Beard 1996), but recent

large phylogenetic analyses (Meng et al. 2003) do not

fully substantiate this (fig. 28.3).

Morphologists, including paleontologists, have extensively

researched relationships within Glires and occasionally

questioned its monophyly (for review, see Meng and

Wyss 2001, Meng et al. 2003). During the last decade, molecular

biologists have challenged both Glires monophyly and

rodent monophyly (Graur et al.1991, 1996, Li et al.1992),

arguing that the guinea pig is more closely related to primates

than to other rodents and that rabbits are more closely related

to primates than to rodents. As pointed out by several

studies, however, early molecular results that indicated a nonmonophyletic

Glires or Rodentia appear to have been artifacts

of small data sets or other methodological problems

(Allard et al.1991, Hasegawa et al.1992, Graur 1993,

Novacek 1993, Catzeflis 1993, Sullivan and Swofford 1997,

Halanych 1998). More recent molecular studies investigating

this claim have included more taxa, more gene sequences,

or both (Madsen et al. 2001, Murphy et al. 2001, 2002,

Waddell et al. 2001). These analyses have corroborated

morphological studies that support Glires monophyly.

Several recent morphological studies continue to support

the monophyly of Glires (Li et al.1987, Novacek 1992a, 1992b,

Luckett and Hartenberger 1993, Shoshani and McKenna 1998,

Meng and Wyss 2001). Figure 28.3 is a phylogenetic tree superimposed

on a stratigraphic distribution of Glires, which

resulted from an analysis of 50 taxa and 227 morphological

characters (Meng et al. 2003). It shows the sister group relationship

of Rodentia and Lagomorpha as determined from

morphological data (for support values, see Meng et al. 2003).

Because of the current topological congruence between molecular

and morphological data, both showing support for

Glires, we anticipate that combined analyses that include fossils

will continue to support this result.

Cetacea

One of the most debated problems in placental mammalian

phylogenetics concerns the position of the order Cetacea

(whales, dolphins, and porpoises). The question of cetacean

affinities is worthy of special consideration here because it

has been examined with diverse data sets, including combined

(total evidence) analyses of multiple genes and morphology.

Most of these studies have focused on identifying

Building the Mammalian Sector of the Tree of Life 499

the extant sister taxon of Cetacea and on determining the

relationship of Cetacea to an extinct group of terrestrial

mammals called Mesonychia, which are hoofed mammals

that are part of the clade Paraxonia. Some of the alternative

hypotheses under consideration include the following: (1)

Cetacea are excluded from a monophyletic Artiodactyla (the

order that includes the even-toed hoofed mammals, e.g., pigs,

hippos, camels, and ruminants) and is the sister group of the

extinct clade Mesonychia, (2) Cetacea are related to a subgroup

of artiodactylans (i.e., hippos) and should be placed

within Artiodactyla (thereby rendering the traditional concept

of Artiodactyla paraphyletic), (3) Cetacea are the sister

taxon of a monophyletic Artiodactyla to the exclusion of

Mesonychia, and (4) Cetacea are the sister taxon of

Mesonychia and this clade is nested within Artiodactyla.

Phylogenetic analyses of Cetacea based on skeletal characters,

anatomy of the digestive tract, transposons, amino

acid sequences, DNA–DNA hybridization scores, and DNA

sequences have been presented recently (reviewed in Gatesy

and O’Leary 2001). Some studies have attempted to include

as much published data as possible, or to qualify their conclusions

based only on partial evidence. Other studies, many

of which produced well-resolved phylogenetic trees, were

obtained only after ignoring potentially contradictory published

evidence (entire character data partitions or parts of

them) and were not, therefore, the product of rigorous phy-

Figure 28.3. Tree of the

relationships of Glires (a

monophyletic clade of rodents

and rabbits) and closely related

taxa: strict consensus of 10 most

parsimonious trees derived from

osteological data (Meng et al.

(2003) plotted against the

stratigraphic record. Thick lines

represent known durations of

fossil lineages. Dashed lines

indicate uncertain durations of

lineages. Daggers indicate

extinct taxa. The polytomy at

the base of Rodentia makes

designation of the membership

of extinct taxa within crown

clade Rodentia somewhat

unclear; taxa such as Paramys,

Reithroparamys, and Cocomys are

not unambiguously part of

crown Rodentia based on these

data. Three crown clades are

defined as follows: Rodentia,

common ancestor of Rattus and

Marmota and all of its descendants;

Lagomorpha, common

ancestor of Ochotona and Lepus

and all of its descendants; Glires,

common ancestor of Ochotona

and Marmota and all of its

descendants. See Meng et al.

(2003) for support values.

Asioryctes

Kennalestes

Barunlestes

Pseudictops

Prolagus

Palaeolagus

Mimotona

Eurymylus

Heomys

Sinomylus

Florentiamys

Cricetops Rattus

Mus

Tataromys

Tsaganomys

Incamys

Cavia

Sciuravus

Cocomys Sciurus

Marmota

Leptictis

Cretaceous Paleogene Neogene Q

Glires

Rodentia Lagomorpha

Plesiadapis

Notharctus

Adapis

Rhynchocyon

Tupaia

Cynocephalus

Petrodromus

Dasyprocta

Paraphiomys

Myocastor

Erethizon

Gregorymys

Paradjidaumo

Tribosphenomys

Paramys

Reithroparamys

Rhombomylus

Matutinia

Mimolagus

Ochotona

Lepus

Anagalopsis

Anagale

Hyopsodus

Phenacodus

Zalambdalestes

Neoreomys

500 The Relationships of Animals: Deuterostomes

logenetic tests. Not surprisingly, the results of these partitioned

analyses have rarely agreed.

One recent example is that of Thewissen et al. (2001).

Although their phylogenetic analysis included previously

undescribed morphological data, their study incorporated

only a subset of previously published data. It excluded some

osteological characters that happened to support a close relationship

between mesonychians and Cetacea to the exclusion

of artiodactylans, the relationship that Thewissen et al.

(2001) claimed to reject. Similarly, Thewissen et al. (2001)

also excluded all published molecular data, thereby barring

molecules from influencing the phylogenetic placement of

Cetacea. Although molecular sequence data have not been

extracted from extinct mesonychians, certain analyses indicate

that the thousands of informative molecular characters

collected to date overwhelmingly contradict the results of

Thewissen et al. (2001) and place Cetacea within a paraphyletic

Artiodactyla, closest to hippopotamids (e.g., Gatesy

et al.1999a, 1999b, Matthee et al. 2001). Exclusion of so

much data in this way makes it impossible to assess the phylogenetic

relevance of this study as presented.

Likewise, many other recent studies of whale phylogeny

included analyses of very small subsets of data in isolation

from the majority of published character evidence. For example,

in a recent study of morphological characters of the

digestive tract, Langer (2001) found that stomach morphology

supported the monophyly of Artiodactyla to the exclusion

of Cetacea. Langer (2001) then asserted that other

morphological characters, which contradicted his hypothesis,

were necessarily the result of convergent adaptation to

an aquatic life and should be dismissed as phylogenetic evidence.

Langer (2001) did not actually include these “convergent”

characters in his matrix, but concluded that

Artiodactyla was monophyletic regardless. Likewise, in a

recent phylogenetic analysis of mtDNA and morphological

data, Luckett and Hong (1998) argued that more than 90%

of the approximately 1000 molecular characters they examined

were too variable to be of any use and were eliminated

from phylogenetic analysis. Putative aquatic adaptations,

such as near-hairlessness in hippos and whales, also were not

considered valid phylogenetic evidence. In this same tradition,

Naylor and Adams (2001) hypothesized that dental

traits, not aquatic specializations or molecular data, were just

too homoplastic or correlated to include in phylogenetic

analysis, leading them to present a preferred tree that excluded

all dental evidence and to propose general arguments

impugning the use of dental characters in mammalian phylogenetics

(see O’Leary et al. 2003).

More inclusive studies of characters and taxa (Gatesy

et al.1999a, 1999b, 2002, O’Leary 2001) have shown that

results based on subsets of the total database are highly unparsimonious

(e.g. O’Leary et al. 2003). We have compiled

two large combined supermatrices of whales, artiodactylans,

and close relatives. The first matrix includes 75 extant taxa

and more than 37,000 characters from three morphological

data sets (Messenger and McGuire 1998, Geisler 2001, Langer

2001), a matrix of SINE transposon insertions (Nikaido et al.

1999, 2001), and 51 genes/gene products from the mitochondrial

and nuclear genomes (fig. 28.4; Gatesy et al. 2002). This

analysis includes most of the characters discussed in Luckett

and Hong (1998), Naylor and Adams (2001), Langer (2001),

and Thewissen et al. (2001) as well as tens of thousands of

characters (mostly molecular) that were published prior to

these papers. The supermatrix of extant taxa did not support

topologies promoted by the authors of partitioned

analyses (fig. 28.4, gray dots) and each of the partitionbased

hypotheses required at least 300 extra character steps

beyond minimum tree length. Some groups supported by

bootstrap scores of 100% and Bremer supports of more than

100 steps were not recovered in the more restricted analyses

of Luckett and Hong (1998), Naylor and Adams (2001),

Langer (2001), or Thewissen et al. (2001).

We also constructed a whale/artiodactylan supermatrix

that included extinct taxa (fig. 28.5). The combined data set

was composed of 50 extinct taxa, 18 extant taxa, and ~36,500

characters. Morphological data were primarily from Geisler

(2001), and molecular characters (including alignment

methodology) came from Gatesy et al. (2002). Per taxon, this

supermatrix has much more missing character data, but should

be a better test of the phylogenetic tree because it includes basal

extinct species, such as primitive whales, early artiodactylans,

and mesonychians. The strict consensus of minimum length

topologies is not well resolved because of the instability of

several taxa, including Mesonychia (fig. 28.5)

Use of a maximum agreement subtree (e.g., Cole and

Hariharan 1996), which summarizes the maximum number

of relationships that are supported by all minimum length

topologies, helps clarify where character conflict is most pronounced.

Instead of collapsing uncertain taxa to basal nodes

as in an Adams (1972) consensus tree, the agreement tree

excludes these taxa and just shows the relationships that are

consistent with all of the equally short trees (fig. 28.6). This

indicates that the differences among the most parsimonious

trees are due to the instability of fossil taxa, not to alternative

relationships for the living taxa (see also discussion of

this in O’Leary 2001). Like the supermatrix for extant taxa

only (fig. 28.4), the combined fossil/extant supermatrix

(fig. 28.5) is consistent with a close relationship between

hippopotamuses and whales, a result that was strictly contradicted

in several analyses that used subsets of published

data (Luckett and Hong 1998, Langer 2001, Thewissen et al.

2001). Furthermore, controversial relationships supported

by the analysis of Naylor and Adams (2001), such as perissodactylan

paraphyly and a grouping of Ovis and Camelus

to the exclusion of Tragulus, were overwhelmingly rejected

(figs. 28.4–28.6). In this analysis the fossil data have not altered

the primary relationships of the extant taxa as determined

from molecular data alone. We do not argue here in

favor of a particular phylogenetic result, but instead suggest

simply that more comparative work will be required to sort

Building the Mammalian Sector of the Tree of Life 501

Figure 28.4. Single minimum length topology of 67,357 steps supported by parsimony analysis of the extant whale–artiodactylan supermatrix with all characters unordered. OTUs are

shown to the right. Higher level taxa are in capitals and are delimited by brackets to the right of OTUs. Data sets are shown at the top of the figure (see for abbreviations, see Gatesy et al.

(2002). Black circles indicate taxa sampled for these data sets; gray represents missing data in the supermatrix. Branch support scores (Bremer 1994) for relationships among cetacean and

artiodactylan families are above internodes. One thousand random taxon addition replicates were used in each constrained heuristic search, but given the complexity of the supermatrix

data set, these branch support scores may be lower than indicated. Bootstrap scores (Felsenstein 1985) that were greater than 69% are indicated below internodes. One thousand bootstrap

replicates were done using heuristic searches of informative characters with simple taxon addition and tree bisection reconnection branch swapping (Swofford 2000). Gray circles at nodes

mark clades that were inconsistent with the combined supertree analysis of Liu et al. (2001) and/or the restricted character analyses of Luckett and Hong (1998), Langer (2001), Naylor

and Adams (2001), or Thewissen et al. (2001). The tree is rooted according to the hypotheses of Madsen et al. (2001) and Murphy et al. (2001).

24

103

97

14

24

22

13

134

98

95

57

8

4

22

6

digestive tract

skeletal + dental

blowhole

cytochrome b

12S rDNA

transposons

CO2

16S rDNA

NADH2

NADH1

CO1

ATP6

CO3

ATP8

NADH4L

NADH3

NADH4

NADH6

NADH5

-hemoglobin

-hemoglobin

-crystallin A

myoglobin

ribonculease

-casein

-casein

Thyroglobulin

MGF

PRKC1

SPTBN1

TNF-A

STAT5

Thyrotropin

IRBP

vWF

-lactalbumin

ADORA3

ADBR2

APP

BDNF

BMI1

CREM

CNR1

PLCB4

EDG1

PNOC

RAG1

TYR

RAG2

ZFX

protamine P1

-fibrinogen

BRCA1

A2AB

ATP7A

97

100

100 100

98

73

86

100

100 99

87

96

100

100

100

100 73

100

100

100

100

100

100

100

100

100

100

100

100

83

100

100

80

100

100

100

100

100

100

78 82

100

100

100

100

100

100

100 100

95

100

100

8

M T nuAA nuDNA mtDNA

Bos sp.

Bubalus bubalis

Bubalus depressicornis

Syncerus caffer

Tragelaphini

Capra hircus

Ovis sp.

Ovibos moschatus

Capricornis crispus

Nemorhaedus sp.

Oryx sp.

Damaliscus sp.

Kobus sp.

Gazella sp.

Aepyceros melampus

Cephalophus sp.

Odocoileus sp.

Cervus sp.

Muntiacus sp.

Alces alces

Giraffa camelopardalis

Okapia johnstoni

Antilocapra americana

Tragulus sp.

Tursiops truncatus

Lagenorhynchus sp.

Globicephala sp.

Orcinus orca

Phocoenidae

Monodon monoceros

Delphinapterus leucas

Inia geoffrensis

Ziphius cavirostris

Mesoplodon sp.

Kogia sp.

Physeter catadon

Balaenoptera physalus

Eschrichtius robustus

Balaenoptera musculus

Balaena mysticetus

Choeropsis liberiensis

Sus scrofa

Babyrousa babyrussa

Tayassu tajacu

Camelus dromedarius

Camelus bactrianus

Lama sp.

Rhinoceros unicornis

Tapirus sp.

Equus sp.

Phocidae

Ailurus fulgens

Ursidae

Procyon lotor

Canidae

Feloidea

Manis sp.

Homo sapiens

Platyrrhini

Leporidae

Rattus norvegicus

Mus sp.

Loxodonta africana

Elephas maximus

Dugong dugon

Trichechus sp.

Procavia capensis

Orycteropus afer

Macroscelidea

Boselaphus tragocamelus

Dicerorhinus sumatrensis

Balaenoptera acutorostrata

Megaptera novaeangliae

Hippopotamus amphibius

Diceros + Ceratotherium

BOVIDAE

CERVIDAE

GIRAFFIDAE

ANTILOCAPRIDAE

TRAGULIDAE

DELPHINIDAE

BALAENIDAE

HIPPOPOTAMIDAE

SUIDAE

TAYASSUIDAE

CAMELIDAE

PERISSODACTYLA

CARNIVORA

PHOLIDOTA

PRIMATES

LAGOMORPHA

RODENTIA

PROBOSCIDEA

SIRENIA

HYRACOIDEA

TUBULIDENTATA

MACROSCELIDEA

PHOCOENIDAE

MONODONTIDAE

INIIDAE

ZIPHIIDAE

PHYSETERIDAE

BALAENOPTERIDAE

+ ESCHRICHTIDAE

ARTIODACTYLANS

CETACEA

ARTIODACTYLANS

OUTGROUPS

502 The Relationships of Animals: Deuterostomes

Figure 28.5. Strict consensus of 4522 minimum length topologies (32,613 steps) supported by

parsimony analysis of the extinct + extant whale supermatrix with all characters unordered [130

random addition replicates in PAUP* beta version 10 (Swofford 2000)]. OTUs are shown to the

right. Higher level taxa are in capitals and are delimited by brackets to the right of OTUs. Data

sets are shown at the top of the figure (M, morphology; T, transposons; nuAA, nuclear amino acid

sequences; nuDNA, nuclear DNA; mtDNA, mitochondrial DNA; for other abbreviations, see

Gatesy et al. (2002), and taxonomic sampling for each data set is indicated by black circles as in

figure 28.4. Gray circles at nodes mark clades that were inconsistent with the combined supertree

analysis of Liu et al. (2001) and/or the analyses of Luckett and Hong (1998), Langer (2001),

Naylor and Adams (2001), and Thewissen et al. (2001). The tree is rooted with Leptictidae (see

Geisler 2001). Daggers indicate fossil taxa. Matrix available through TreeBASE.

skeletal + dental

digestive tract

blowhole

cytochrome b

12S rDNA

transposons

CO2

16S rDNA

NADH2

NADH1

CO1

ATP6

CO3

ATP8

NADH4L

NADH3

NADH4

NADH6

NADH5

-casein

-casein

Thyroglobulin

MGF

PRKC1

SPTBN1

TNF-A

STAT5

Thyrotropin

IRBP

vWF

-lactalbumin

ADORA3

ADBR2

APP

BDNF

BMI1

CREM

CNR1

PLCB4

EDG1

PNOC

RAG1

TYR

RAG2

ZFX

protamine P1

-fibrinogen

BRCA1

A2AB

ATP7A

M T nuDNA mtDNA

Odocoileus

Bos

Ovis

Tragulus

Cainotherium

Leptomeryx

Hypertragulus

Balaenoptera

Tursiops

Delphinapterus

Physeter

Georgiacetus

Protocetus

Basilosaurus

Remingtonocetus

Ambulocetus

Pakicetus

Hippopotamus

Choeropsis

Cebochoerus

Sus

Tayassu

Entelodontidae

Perchoerus

Elomeryx

Homacodon

Gobiohyus

Lama

Camelus

Poebrotherium

Eotylops

Diacodexis pakistanensis

Wasatch Diacodexis

Bunomeryx

Agriochoerus

Merycoidodon

Mixtotherium

Xiphodon

Amphimeryx

Leptoreodon

Heteromeryx

Protoceras

Equus

Mesohippus

Heptodon

Hyracotherium

Hyopsodus

Phenacodus

Meniscotherium

Canis

Rattus

Vulpavus

Arctocyon

Eoconodon

Andrewsarchus

Hapalodectes hetangensis

Hapalodectes leptognathus

Dissacus praenuntius

Dissacus navajovious

Mongolian Dissacus

Sinonyx

Pachyaena gigantea

Pachyaena ossifraga

Mesonyx

Synoplotherium

Harpagolestes

Orycteropus

Leptictidae

ARTIODACTYLANS

CETACEA

ARTIODACTYLANS

PERISSODACTYLA

MESONYCHIA

CARNIVORA

Building the Mammalian Sector of the Tree of Life 503

paraphyly. Importantly, however, Afrotheria was not supported

in the parsimony analysis.

Liu et al. (2001) published a supertree analysis, which

also resulted in a large mammalian tree, with 91 terminal taxa.

This combined summary of previous morphological and

molecular studies was largely congruent with traditional

hypotheses of relationship based on morphology. In other

words, the most basal clade was an edentate group, and the

monophyly of Insectivora, Artiodactyla, Rodentia, and Glires

was supported, but the monophyly of Afrotheria and

Laurasiatheria was not. Liu et al. (2001) attempted to limit

the overall redundancy of information in their supertree data

set by using only the most recent and comprehensive published

analysis for each gene, but there were still considerable

duplications of evidence in the supertree data set.

Reviews and assumptions of monophyly that were not based

on primary data analysis also were included as evidence in

out all the relationships among living whales and their extinct

relatives.

Mammalian Supertrees and Supermatrices

A number of recently published large-scale molecule-based

supermatrix analyses include increasingly greater numbers

of taxa (>50) analyzed simultaneously. For example, Murphy

et al. (2001) performed a simultaneous analysis of 64 mammal

taxa using data from 18 different gene segments. Their

results showed some variance in tree topology depending on

the method of phylogenetic analysis (e.g., parsimony vs.

maximum likelihood). These authors figured the maximumlikelihood

tree, which showed the clades Glires, Xenarthra

(sloths, anteaters, and armadillos), Afrotheria, and Laurasiatheria.

The tree also supported results like artiodactylan

Figure 28.6. A maximum agreement subtree (Cole and Hariharan 1996) of shortest topologies

found for the extinct + extant whale supermatrix (see strict consensus in fig. 28.5). This shows

relationships that are stable among all most parsimonious trees. The phylogenetic positions of

Mesonychia and some other taxa vary among minimum length trees and are therefore excluded

from the agreement subtree. Gray circles at nodes mark relationships that are inconsistent with

the combined supertree analysis of Liu et al. (2001) and/or the analyses of Luckett and Hong

(1998), Langer (2001), Naylor and Adams (2001), and Thewissen et al. (2001). Daggers indicate

fossil taxa.

ARTIODACTYLANS

CETACEA

ARTIODACTYLANS

PERISSODACTYLA

CARNIVORA

Odocoileus

Bos

Ovis

Tragulus

Cainotherium

Leptomeryx

Hypertragulus

Protoceras

Agriochoerus

Merycoidodon

Balaenoptera

Tursiops

Delphinapterus

Georgiacetus

Protocetus

Remingtonocetus

Ambulocetus

Pakicetus

Hippopotamus

Choeropsis

Cebochoerus

Bunomeryx

Sus

Tayassu

Entelodontidae

Elomeryx

Homacodon

Lama

Camelus

Poebrotherium

Eotylops

Xiphodon

Amphimeryx

Diacodexis pakistanensis

Wasatch Diacodexis

Equus

Mesohippus

Heptodon

Hyracotherium

Eoconodon

Andrewsarchus

Vulpavus

Orycteropus

Leptictidae

TUBULIDENTATA

504 The Relationships of Animals: Deuterostomes

the supertree data set. Gatesy et al. (2002) suggested that

these duplications of evidence and other problems with

supertree analysis led to phylogenetic results that were not

supported by the underlying character data. Actual analysis

of the characters (fig. 28.4) by those authors shows that just

within Paraxonia (whales + artiodactylans), the supertree

topology is more than 450 character steps less parsimonious

than the minimum length tree supported by the data (i.e.,

the supermatrix analysis). The reanalysis by Gatesy et al.

(2002) also supports monophyly of Rodentia and Glires,

albeit with the minimum sample size.

Applying the Phylogeny of Mammals

to the Determination of the Age of a Clade:

Ghost Lineages and Molecular Clocks

Great interest has been focused on determining divergence

times of different mammal clades and answering such questions

as what is the oldest placental mammal, the age of the

clade Placentalia (e.g., the basal split within Placentalia), and

the age of the ancestral eutherian lineage leading to Placentalia.

Calculating these dates is fundamentally related to

phylogeny reconstruction. For familiar clades (like Placentalia)

whose names have been in circulation under a variety

of definitions, it is particularly important to employ explicit

clade definitions (e.g., crown clades) when comparing divergence

dates derived exclusively from fossil evidence with

those derived from calibrated molecular evidence. Obviously

(with the exception of occasional ancient DNA discoveries)

the only divergence times that can be estimated using calibrated

molecule-based divergence times are those between

pairs of extant clades. Many other divergence dates can be

assessed using the fossil record alone (e.g., the divergence of

two extinct clades, the divergence of one extinct and one

extant clade), but these cannot be compared directly with

molecule-based divergence dates. It is key to compare explicit

clade branching points regardless of the method of determining

the dates.

One means of determining the age of a divergence time

that relies on few assumptions is to compare the clade in

question with the age of its sister taxon. This process was

described by Marshall (1990) and formalized by Norell

(1992) as ghost lineage analysis. Ghost lineage analysis entails

simply the assumptions of phylogeny reconstruction.

Ghost lineages are predicated on the idea that, if two monophyletic

taxa are sisters, then they must have split at the same

time. The oldest species among the taxa in either clade puts

a minimum age on the split (Marshall 1990; fig. 28.7). For

example, in figure 28.7, the split between clade B and clade

A marks the most basal branching point within crown clade

Placentalia (as defined above). To refer to a recent analysis

(Springer et al. 2003), this would represent the split between

the clade (Xenarthra + Boreoeutheria) and Afrotheria, for

example. The oldest of these clades is B making it both the

oldest placental and the clade that puts a minimum date of

40 Myr (million years) on the basal split within Placentalia

in this hypothetical example. The segments of time for the

lineage that are not recorded by fossils but which are dictated

by the phylogeny are referred to as ghost lineages

(fig. 28.7, dashed lines). Clade A has a ghost lineage of 20

Myr, during which time it had already split from B (but no

fossils of clade A have been found in this interval). The split

between Placentalia and its extinct eutherian sister taxon can

also be calculated. If clade C is the eutherian sister taxon of

Placentalia and is 90 Myr old, then ghost lineage logic dictates

that the ancestral eutherian lineage leading to Placentalia

must have split from clade C at the same time [90 million

years ago (Mya)]. Ghost lineages can be calculated on any

cladogram, even a cladogram of extant taxa alone. They are

most effective, however, if the fossils that are part of the clade

have been analyzed simultaneously with the extant taxa. If

an older member of the clade is found, the phylogenetic

analysis and the hypothesis of the age of the clade can be

revised accordingly. Using a crown clade definition, the minimum

age estimate for the origin of Placentalia is synonymous

with the timing of the first split within Placentalia. The oldest

species nested within crown clade Placentalia will determine

the age of this divergence.

An alternative means of calculating the age of a clade is

to use a molecular clock. This generally has been described

Figure 28.7. Schematic explaining the ghost lineage concept and

how it can be applied to calculating the date of the basal split

within the crown clade Placentalia and the age of the ancestral

eutherian lineage leading to Placentalia using a hypothetical

example. For the two members of Placentalia, clade A is younger

than its sister, clade B. Although clade A’s actual fossil record

extends to only 20 Mya, clade A must have split from B at least 40

Myr old based on its phylogenetic relationships and the age of

clade B. In other words, taxon A has a ghost lineage (dashed line 1)

of 20 Myr. Because clade B is the oldest member of Placentalia its

age puts a minimum divergence date on the basal split within

Placentalia. This is the relevant date for comparison with moleculebased

estimates of the origin of Placentalia. The sister taxon of

Placentalia (clade C) is older than either taxon B or taxon A; therefore,

there is a ghost lineage of 50 Myr (dashed line 2) extending

the date of the ancestral eutherian lineage leading to Placentalia.

clade B clade A

clade C

Placentalia

50 mya

Recent

100 mya

1

2

Building the Mammalian Sector of the Tree of Life 505

as “an independent means of estimating times of origin for

extinct clades” (Smith and Peterson 2002: 66) relative to the

use of paleontological data. In its original formulation the

molecular clock model of uniform rates of gene sequence

change (Zuckerkandl and Pauling 1962, 1965) was adopted

as a means of determining the absolute age of the divergence

event between two lineages given a certain calibration

(fig. 28.8). This differs from ghost lineages, which amount

to a minimum estimate of divergence. The rate of divergence

(e.g., the number of nucleotide changes that occurred in a

given lineage since a splitting event) is not, however, a known

quantity. The rate is derived from independent evidence used

to calibrate the clock. Sometimes the calibration is a date of

a selected fossil or fossils. Alternatively the date of a major

geological event, such as the opening of the Atlantic or the

separation of South America and Africa, has been equated

with the time of separation of two taxa (reviewed in Smith

and Peterson 2002). Once the calibration is established and

the rate of divergence calculated, that rate is assumed to be

accurate for all lineages compared (fig. 28.8). Figure 28.8

illustrates the basic equation for the calculation of divergence

times using a molecular clock. This simple formula

has been applied in numerous cases, but there have been

criticisms of the very rationale for even applying the molecular

clock (Novacek 1982, Goodman et al.1982, Ayala

1997, Ayala et al.1998). In its simplest incarnation, the

molecular clock has been largely “discredited” (Smith and

Peterson 2002).

Unlike ghost lineage analyses, molecular clock analyses

entail not only the assumptions of phylogenetic analysis, but

also at least two other important additional assumptions: (1)

that the dates of origin of the fossils used to calibrate the rate

of gene change (“clock”) are accurate, and (2) that nucleotide

changes (substitutions) occur at a uniform rate (fig. 28.8B;

Zuckerkandl and Pauling 1965; see also Li 1997) or some

“relaxation” of rate uniformity (methods reviewed in Smith

and Peterson 2002: 75). Each of these assumptions introduces

a separate set of problems.

Regarding calibration, Novacek (1982) noted that any

error in the calibration of a divergence taken from a relevant

fossil taxon (T in fig. 28.8) could grossly affect the estimate

of divergence dates based on the clock model, a problem

rediscovered by Lee (1999) and Alroy (1999). For example,

molecular estimates of divergence are typically calibrated

simply by a fossil’s first appearance; however, this could

be an underestimate of age if a well-tested phylogenetic

hypothesis has not been taken into consideration. It would

be appropriate to check the age of any taxon used as a calibration

point against the age of its sister to see if the

calibration point has a ghost lineage extending its age in

time. This type of calibration has rarely been explicitly

employed.

Second, and equally important, the rate (and thereby the

date of a split between taxa) will also be miscalculated if a

given gene has evolved at a faster rate in one of the two taxa

Figure 28.8. In its simplest formulation (average distance

methods) a molecular clock is calibrated on the basis of the split

between two taxa, for example, taxon W and taxon X (A), that is

assumed to have occurred at a given date T. Using the number

of nucleotide differences between X and W (e.g., 50 bp), the

rate of nucleotide substitution, K, for other taxa either in the

clade or outside of it can be calculated using the formula shown

(C). Once this rate is established, the time elapsed between the

split of taxon D and taxon G (tDG), for example, can be calculated.

In both the initial calculation of rate and in the calculation

of the split between D and G, the assumption is that the rate of

change is distributed equally down each lineage (B). It is

entirely possible, however, that this assumption is violated such

as is shown in (C), even for closely related taxa.

K = number of nucleotide

substitutions between

two taxa

r = rate of substitution

T = calibration time

t = calculated time of split

between two taxa (D,G)

W C D E F G

W X

W

X

T

1

2 K 1

2 K

3

4 K

1

4 K

r =

K

2T

tDG =

2r

KDG

A

B

C

X

T

506 The Relationships of Animals: Deuterostomes

than in the other (fig. 28.8C). Enthusiasts of the molecular

clock have responded to this criticism by abandoning the

clock in a strict sense for a variety of different types of molecular

estimates of divergence. These have been argued to

be more robust because they either assess rate heterogeneity

a priori or because they have a built-in ability to account for

different rates of nucleotide evolution among taxa.

For example, investigators have applied relative-rate tests

(Sarich and Wilson 1967; see also Tajima 1993) to compare

the rates of substitution in a set of taxa, rejecting those genes

that show significant rate heterogeneity. Many investigators

(e.g., Kumar and Hedges 1998) have then gone on to apply

the clock on genes that do not possess significant rate heterogeneity.

Wu and Li (1985) used a relative rate test to compare

rodent and human lineages using either artiodactylans

or carnivorans as outside reference taxa. They concluded that

the rate of synonymous substitution is about twice as high

in the rodent lineage as in the human lineage. This did not

prompt the authors to reject the molecular clock; however,

instead they suggested that differences in rates could be tied

to various biological parameters, arguing in this case that

rodents with their short generation times would be expected

to share a fairly uniform rate, but one much higher than that

of humans and other primates. Similarly, Li and Graur (1991:

85) stated, “Although there is no global clock for the mammals,

local clocks may exist for many groups of closely related

species.” There is, however, no particular evidence that

gene rates are necessarily less heterogeneous between closely

related taxa than between distantly related taxa, a matter that

detracts from any justification for a distinction between “local”

and “global” clocks. Furthermore, as discussed by Ayala

et al. (1998) and Smith and Peterson (2002: 73), relative rate

tests are not very powerful because they allow “considerable

rate variation to go undetected” and “predicted times of origin

[to be] wrong by as much as 50%” depending on the

amount of unperceived rate heterogeneity. Thus, rate heterogeneity

in molecular estimates of divergence times remains

an important problem.

Complicating matters has been the observation that different

genes also often provide different estimates of divergence.

Hence, a number of workers have tried to correct the

problem of rate heterogeneity by simply sampling more genes

for the split in question and averaging the results. Here, it is

argued that the large errors associated with estimating divergences

based on the clock model can be minimized by incorporating

data from many nucleotides in several genes

(Fitch 1977, Li and Graur 1991). Such approaches draw on

large sample sizes of sequence information and have been

applied to estimates of divergence dates of many mammal

and bird lineages (Hedges et al.1996, Kumar and Hedges

1998). These efforts to broaden nucleotide sampling in order

to achieve supposedly more reliable estimates address

only one dimension of uncertainty associated with these

calculations—the possible heterogeneity in rates for different

genes. They do not correct for the above noted problem

of variation in rates that exist between two lineages after a

given splitting event (fig. 28.8C).

Modeling rate variation has become an alternative to the

methods above as reviewed in Smith and Peterson (2002; see

also Cutler 2000). Collectively these methods relax the strict

assumption that rates of molecular evolution stay the same

over time. The accuracy of the estimate of a divergence, however,

depends on the reliability of the model of molecular

substitution, which can be problematic given that the “actual

patterns of amino acid or nucleotide substitution . . . are

usually unknown” (Smith and Peterson 2002: 75). For example,

the assumption that closely related species have similar

rates of evolution described above, has found its way into

certain model-based estimates of molecular divergence,

where this is referred to as autocorrelation of rates (e.g.,

Sanderson 1997, Thorne et al.1998). However, as stated

above, it is unclear that there is broad-based empirical evidence

supporting this claim. Furthermore, model-based

methods typically require a tree a priori because these methods

require comparisons to be made topologically. Typically,

these trees are derived not from combined data analyses but

instead exclusively from molecular data. This tendency introduces

potential shortcomings because it ignores the impact

of nonmolecular data on the topology.

Divergence Times for Placentalia

Dating the radiation of Placentalia has become one of the

most discussed topics in mammalian phylogenetics, in part

because it has been promoted as a notorious “molecules versus

morphology” debate in the scientific press. As noted

above, the discussions have been complicated by pronounced

variation in the definitions of such terms as “eutherian,” “placental,”

and “therian” (e.g., compare Novacek 1999, Ji et al.

2002, Luo et al. 2002, Smith and Peterson 2002). Here we

employ the stem and crown clade definitions outlined above

(fig. 28.2) to explain the dating of clades using ghost lineages

and as a basis for supporting our best assessment of the minimum

ages of certain clades based on ghost lineages.

Calculating the minimum estimate for the age of the basal

split within Placentalia using ghost lineages requires a tree

that is a well-tested phylogenetic hypothesis of placental relationships

that includes living and fossil species, in particular,

fossil species from the Cretaceous. This permits discovery

of the result that Cretaceous taxa belong within Placentalia.

Preferably this tree would be derived from a combined (simultaneous)

analysis of different data types (e.g., molecular

and morphological) for both extant and extinct taxa. Global

analyses of this kind for Placentalia, however, are only currently

underway. For the purposes of illustrating the ghost

lineage method, we discuss here how such a minimum age

would be calculated using results from smaller phylogenetic

analyses of Placentalia. Any minimum age calculations presented

here would be open to testing by larger, more diverse

total evidence analyses.

Building the Mammalian Sector of the Tree of Life 507

Relevant analyses in the literature fall into two groups: (1)

those that test the relationships of a number of extant placental

taxa (more than two OTUs) and one or more extinct Cretaceous

taxa, and (2) those that sample one representative extant

placental crown clade member (OTU) and several extinct taxa

from the Cretaceous. The second type of analysis obviously does

not contain enough placental taxa (more than one) to permit

the discovery of a Cretaceous taxon within Placentalia, but these

types of analyses do contribute some information on the distribution

of Cretaceous taxa within Theria (fig. 28.2).

Example analyses that fall into the first category are

O’Leary and Geisler (1999) for Paraxonia (see also O’Leary

Figure 28.9. Strict consensus of eight minimum length

topologies (180 steps) derived from a matrix originally analyzed

by Novacek (1992a) with several taxa (e.g., Ukhaatherium) and

characters added. Cretaceous taxa (Zalambdalestes,

Uhkaatherium, Kennalestes, and Asioryctes) all fall outside of the

crown clade Placentalia as its sister taxon. Fossil taxa included

within Placentalia (e.g., Palaeoryctes) date to the early Paleocene.

Using ghost lineages, this places an early Paleocene minimum

divergence data on the basal split within Placentalia. Although

the extinct taxa Plesiadapis and Anagale form part of a polytomy

with other clades in Placentalia, in each minimum length

topology these taxa fall within crown Placentalia and are

therefore labeled accordingly here. The parsimony search

(PAUP*, ver. 4.10) was heuristic with tree bisection and

reconnection branch swapping, amb-option (internal branches

collapsed if the minimal possible length of the branch is zero) in

effect, multistate taxa treated as a polymorphism, all characters

unordered (1000 random addition replicates). Tree rooted

through Monotremata; images as in figure 28.1. Daggers

indicate wholly extinct taxa; taxa without daggers also often

contain a majority of extinct species. Consistency index =

0.6667; homoplasy index = 0.3333 (both of the former

excluding uninformative); retention index = 0.7447; rescaled

consistency index = 0.5151. Matrix available through TreeBASE.

Marsupialia

Vincelestes

Zalambdalestes

Kennalestes

Ukhaatherium

Asioryctes

Plesiadapis

Macroscelidea

Lagomorpha

Rodentia

Palaeanodon

Carnivora

Microsyops

Palaeoryctes

Sirenia

Desmostylia

Proboscidea

Placentalia

Stem

Eutheria

Monotremata

Anagale

Dermoptera

Chiroptera

Primates

Scandentia

Metacheiromys

Xenarthra

Manis

Orycteropus

Insectivora

Leptictidae

Artiodactyla

Cetacea

Hyracoidea

Perissodactyla

and Uhen 1999), Meng et al. (2003; see also fig. 28.3) for

Glires, and an updated version of Novacek (1992a; fig. 28.9;

see also Novacek 1999) that includes a number of newly

discovered taxa [Shoshani and McKenna (1998) does not fit

this category because it does not treat Cretaceous taxa as

OTUs]. Inspection of the trees noted above for Glires and

Paraxonia shows that the Cretaceous taxa included in each

case fall outside the branching points between sampled members

of Placentalia. An analysis across Placentalia (fig. 28.9)

that includes the recently discovered and highly complete

taxon Ukhaatherium (Novacek et al.1997) indicates that the

Cretaceous taxa (Zalambdalestes, Uhkaatherium, Kennalestes,

508 The Relationships of Animals: Deuterostomes

and Asioryctes) all fall outside the basal divergence within

Placentalia. These taxa form a eutherian sister clade to

Placentalia. This analysis overturns previous hypotheses

that these Cretaceous taxa belonged within Placentalia (e.g.,

Novacek 1992b; see also Novacek et al.1997, Novacek

1999). Thus, based on this analysis, the basal split (here a

polytomy; fig. 28.10) within Placentalia (fig. 28.10, position

1) is determined by the oldest taxon in the clade, Palaeoryctes,

which dates to the Early Tertiary (specifically the Early Paleocene,

~64 Mya).

These ghost lineage calculations for minimum divergence

times can be compared with recent molecular clock estimates.

Sequence data representing many loci have been used under

a clock assumption to determine molecular estimates of

divergence for vertebrate groups, including placentals (e.g.,

Hedges et al.1996, Springer 1997, Kumar and Hedges 1998,

Hedges and Poling 1999). Molecular estimates for the origin

of placental clades have often been markedly older than

those suggested by most calculations based on the fossil

record. Hedges et al. (1996), for example, showed dates of

more than 100 Myr for the origin of several lineages within

crown Placentalia (e.g., primates, edentates, rodents, and

artiodactylans). Because these clades are within Placentalia,

these early dates, if corroborated, could pull back the dates

of origin for many placental clades (depending on tree topology)

well into the Cretaceous.

A second study by Kumar and Hedges (1998) greatly

expanded coverage to 658 genes representing 207 vertebrate

species and showed similarly ancient divergences, including

a divergence time of 129 Myr, for certain members of

Placentalia. These authors also estimated the split between

Marsupialia and Placentalia to have occurred 173 Mya.

Kumar and Hedges’s (1998) analysis of mammalian divergence

times has been revised, most notably by Eizirik et al.

(2001), who analyzed 10,000 base pairs (bp) in 64 mammal

taxa to arrive at somewhat more recent divergence times for

most mammal orders, between 64 and 109 Myr, estimates

that conformed more closely with those of Springer (1997).

Most recently Springer et al. (2003) also contributed new

dates based on 19 nuclear and three mitochondrial genes.

These analyses employed model-based molecular estimates

of divergence. In particular, the model of Springer et al.

(2003) incorporated multiple fossil constraints on divergence

times and allowed rates of molecular evolution to vary

on different branches. They still obtained the result that not

only were there several supraordinal divergences within the

Cretaceous but also, and importantly, divergences within four

placental orders (Lipotyphyla, Rodentia, Primates, and

Xenarthra) occurred prior to the Cretaceous–Tertiary

boundary, as early as 74–77 Mya. Their estimate for the age

of the basal split within Placentalia was 97–122 Mya. Clearly,

these results disagree with the numbers presented above

derived from ghost lineage calculations, which put the age

of the basal split within Placentalia in the Early Tertiary at

~64 Mya. The fossil record (fig. 28.3) shows no evidence of

Figure 28.10. Schematic describing the age of the clade

Placentalia and closely related clades (daggers indicate fossil

taxa). (1) indicates a fossil taxon that falls within the crown

clade Placentalia. (2) indicates a fossil that is the eutherian sister

taxon of Placentalia. (3) indicates a fossil that is another

Eutherian stem taxon to Placentalia (in this case the oldest

member of Theria). Fossils that fall at position 2 or 3 are not

directly relevant to calculating the minimum age of Placentalia.

Current ghost lineage calculations indicate that all fossils within

Placentalia (position 1) have a minimum age of approximately

64 Myr old (fig. 28.9). Taxa in position 2 have a minimum age

of Late Cretaceous (77 Myr old; see fig. 28.9; see also Rougier

et al. 1998, Rauhut et al. 2002; or 65 Myr old based on

Protungulatum, Ji et al. 2002); taxa in position 3 have a minimum

age of 125 Myr (based on Eomaia, Ji et al. 2002); neither

of these is directly relevant to the age of Placentalia. Taxa in

position 4 [Pucadelphys (Ji et al. 2002) or Andinodelphys (Rougier

et al. 1998)] put a minimum age of Early Paleocene on the

ancestral metatherian lineage leading to Marsupialia, and those

in position 5 on the oldest member of Metatheria [Early

Cretaceous based on Kokopelia, Ji et al. (2002) and Luo et al.

(2002)]. Eomaia, if correctly dated at 125 Myr old, currently

qualifies as the oldest known member of the crown clade

Theria, but contra Ji et al. (2002), its discovery does not

promote congruence between molecular and paleontological

estimates of the basal split within Placentalia. The outgroup to

Theria, position 6 (Vincelestes, fig. 28.9; Slaughteria or

Pappotherium, Rougier et al. 1998; or Kielantherium, Rauhut

et al. 2002, Ji et al. 2002) also dates to the Early Cretaceous or

possibly Late Jurassic (Peramus; see Ji et al. 2002).

Marsupialia

1

3

Metatheria

Theria

Quaternary

Tertiary

Late

Cretaceous

Mesozoic Era Cenozoic Era

2 mya

65 mya

144 mya

Eutheria

Placentalia

Early

Cretaceous

98 mya

6

4

Recent

5

2

Building the Mammalian Sector of the Tree of Life 509

characters, even though the authors themselves included

many dental characters in their matrix that show homoplasy

on their most parsimonious trees. A priori elimination of data

that might produce a conflicting result is clearly not justified

if the goal is to provide a robust test of alternative hypotheses

of relationship (see similar problems in the above

discussion on cetacean evolution).

Recent studies of some of the above taxa further suggest

that Cretaceous forms such as zalambdalestids are stem

groups outside crown placentals. When Meng et al. (2003)

included Archibald et al.’s (2001) surrogate crown Glires and

zalambdalestids in an extensive cladistic analysis of Glires,

zalambdalestids did not emerge as a member of crown Glires

or as its sister clade (fig. 28.3). Instead, zalambdalestids occupied

a very basal position on the tree several nodes away

from the other crown placental taxa in that analysis (e.g.,

tupaids, dermopterans, and macroscelideans). Wible et al.

(2004), in the most comprehensive comparisons to date of

zalambdalestid morphology, also found no clear evidence for

a close affinity between zalambdalestids and Glires, or between

zalambdalestids and another placental subclade.

A second paper claimed an emerging congruence between

molecular and paleontological estimates of diversification

for Placentalia; Ji et al. (2002: 816) identified a 125

Myr-old skeleton, Eomaia, from Northern China as a “eutherian

(placental)” mammal and suggested that this discovery

indicated a much more ancient date of origin for Placentalia

than had been demonstrated previously from fossil evidence.

The phylogenetic analysis in Ji et al. (2002), however, shows

Eomaia several branches outside the basal split within

Placentalia (fig. 28.10, position 3). Topologically, Eomaia is

actually a very basal member of Eutheria on the stem to, and

well outside of, Placentalia. Eomaia is of no direct relevance

to molecular estimates of the basal split within Placentalia.

Even if Ji et al 2002) were to use a stem-based definition of

Placentalia, Eomaia would still not be relevant to the controversial

molecule-based estimates of divergence within

Placentalia. This is because the molecule-based estimates

apply to the basal split within Placentalia and topologically

Eomaia is far removed from that split.

Moreover, Ji et al. (2002) failed to point out a more relevant

implication of the age of Eomaia—that, if correctly

dated, it is one of the oldest known members of the crown

clade Theria (figs. 28.2, 28.10). It provides paleontological

evidence that the split between Marsupialia and Placentalia

is at least 125 Myr old, a date that is still 50 Myr more recent

than the estimate of 173 Myr for this divergence which

emerged from the molecular clock analysis of Kumar and

Hedges (1998). Contra Kumar and Hedges (1998: 917), it

is not the case that “the molecular estimate for the marsupial-

placental split, 173 Myr ago, corresponds well with the

fossil based estimate (178–143 Myr ago).” Their characterization

of the fossil-based estimate is too ancient and is not

supported by ghost lineage analysis.

Clearly there remains a marked lack of agreement bea

split within Rodentia on the order of 74 Mya or within

Glires at greater than 80 Mya to match the ages of clade diversifications

in Springer et al. (2003).

A number of published analyses or remarks suggest otherwise,

that there is growing consensus between paleontological

and molecular estimates for divergences within Placentalia

that occurred well within the Cretaceous. For example,

Springer et al. (2003: 1060) argued that “McKenna and Bell

[1997] . . . recognized 22 genera from the Late Cretaceous

and one genus from the Early Cretaceous as crown-group

Placentalia,” which would seem to lend paleontological support

for results in Springer et al. (2003). The McKenna and

Bell (1997) classification, which is a monumental literature

review and synthesis of taxonomic work on Mammalia, is not,

however, a classification based on a phylogenetic analysis. Strict

phylogenetic readings of this classification may result in claims

that are not necessarily based on analysis of character data.

Archibald (1996) and Archibald et al. (2001) also argued

for the antiquity of some lineages within Placentalia based

on new fossils known as zhelestids and zalambdalestids from

Uzbekistan. These fossils are thought to be between 85 and

90 Myr old, and using a cladistic analysis of dental, jaw, and

snout characters, these authors concluded that zhelestids

were early members of a “superorder” of placental ungulates

(hoofed mammals) and that zalambdalestids were associated

with Glires. In other words, these authors hypothesized that

their Cretaceous fossils fell in position 1 in the schematic in

figure 28.10, within crown clade Placentalia. If supported,

these proposals would obviously offer paleontological evidence

for a much earlier origin of certain placental clades and,

using ghost lineages, for the clade Placentalia as a whole.

Because of the taxon sampling, however, the Archibald

et al. (2001) analysis did not amount to an explicit test of

the affiliation of the new fossil taxa with Placentalia. Archibald

et al. (2001) analyzed four extinct taxa (the Tertiary ungulates

Protungulatum and Oxyprimus and the Tertiary Glires

taxa Tribosphenomys and Mimotoma) that they argued were

representative crown placentals. Although Tribosphenomys

and Mimotona have subsequently been demonstrated to be

members of Glires (Meng et al. 2003), and thus Placentalia,

this is not the case for Protungulatum. Furthermore, ungulate

phylogeny in general is very much in flux [e.g., compare

Novacek (1992a, 1992b) with Springer et al. (2003) or

Gatesy et al. (2002)]. Accordingly, robust tests of membership

within any crown clade should include living members

of that clade in the analysis.

Character sampling was also problematic in Archibald

et al. (2001). Cranial and postcranial characters cited as evidence

of the monophyly of Placentalia to the exclusion of

forms like the zalambdalestids (Novacek et al.1997) were not

considered. Instead of incorporating these characters into

their data matrix, Archibald et al. (2001) excluded them,

arguing that they did not occur universally within placentals.

Such an operation implicitly suggests that some characters

that may show homoplasy are more expendable than other

510 The Relationships of Animals: Deuterostomes

tween minimum ages for the origin of Placentalia (and clades

within it) as calculated using ghost lineages and dates of divergence

derived from molecules. Our observations here

corroborate those of Rodrнguez-Trelles et al. (2002: 8112),

who noted that “although data sets have become larger and

methods of analysis considerably more sophisticated, the

discrepancy between the fossil record and molecular dates

has not disappeared.” Indeed, several other problems with

the molecular estimates can be noted. For example, the more

conventional clock method employed by Kumar and Hedges

(1998) does not actually require an a priori tree as some

molecular estimates of divergence do. However, their results

imply topologies some of which are discrepant with published

trees based on character data (morphological, molecular

or combined). The most conspicuous example is Glires

(fig. 28.11), a clade that has been shown on the basis of

morphological and molecular data to be monophyletic with

respect to humans (e.g., Meng et al. 2003, Gatesy et al. 2002;

see also figs. 28.3, 28.4). The topology implied by the Kumar

and Hedges (1998) analysis is incongruent with the topologies

of character-based analyses, because if rodents and rabbits

are more closely related to each other than either is to

humans, then the clock estimates should show humans splitting

from rodents and rabbits at the same time. Similarly, the

implied topology of Kumar and Hedges (1998) for ruminants,

suids, and cetaceans (using the mean as an indicator

of the sequence of divergence) is not consistent with published

parsimony analyses based either on molecules, morphology,

or both (see Gatesy and O’Leary 2001). Finally, the

sequence of divergence of Paraxonia (sometimes referred to

as Cetartiodactyla), Carnivora, and Perissodactyla is not corroborated

by molecule-based phylogenetic analyses (Gatesy

et al.1999a, 1999b, Murphy et al. 2001) or supermatrices

(Gatesy et al. 2002). Presented with these conflicting results,

we place greater importance on the tree topology because it

introduces fewer assumptions.

Certain recent authors (Smith and Peterson 2002,

Springer et al. 2003) have argued that they find “convincing”

(Smith and Peterson 2002: 82) the variety of molecular

estimates, including “linearized tree methods that assume a

single rate, quartet dating methods allowing two rates, and

new Bayesian methods that allow rate variation across the

topology” (Springer et al. 2003: 1061), because they all produce

some intraordinal Cretaceous divergence dates for

Placentalia. Less sanguine, however, are the observations of

Rodrнguez-Trelles et al. (2001, 2002). Rodrнguez-Trelles et al.

(2002: 8114) described a fundamental flaw inherent in clockbased

estimates of divergence that “leads to dates that are

systematically biased toward substantial overestimation of

evolutionary times,” especially with large samples of molecular

sequence data. This is extremely problematic for molecular

estimates of divergence, because large sample sizes were expected

to improve these estimates. It remains unconvincing

that the explanation for the incongruence between molecular

and paleontological estimates is simply a poor fossil record.

Nonetheless, Smith and Peterson (2002: 65) insisted repeatedly

that “a global rock bias” exists because “paleontological

sampling in the Late Cretaceous is still too restricted geographically

to draw any firm conclusions about the existence of a Pre-

Tertiary record for modern orders [i.e., Placentalia].” But these

authors did not address the arguments of Novacek (1999: 246),

who noted that despite persistent geographic irregularities in

the mammal fossil record, it remains “much enriched and much

studied compared to other vertebrate groups” with many taxa

documented from both the Cretaceous and the Tertiary periods.

He argued that “[a]pparent patterns of mammalian distribution

are not so easily ascribed to biases due to an impoverished

record, as they might be for birds, amphibians, or other groups.”

This argument is consistent with the results of Foote et al.

(1999), who showed that it was extremely unlikely statistically

that members of Placentalia existed in the Cretaceous but simply

have not been found as fossils. Smith and Peterson (2002:

71) doubted the Foote et al. (1999) results, based largely on

North American and some Asian localities, could be generalized

globally, because several “molecular phylogenies suggest

a Gondwanan origin for many mammalian orders.” The idea

of a Gondwanan origin for Placentalia, however, remains untested

by morphology or combined analyses and may not be

substantiated once ancient placental fossils have been analyzed

simultaneously with molecular sequences in phylogenetic

analyses.

Thus, we fail to see why a convergence among molecular

methods is a compelling validation of their results. All of

these studies share the same premise that assessment of a

large number of nucleotides somehow increases reliability of

the molecular dates, but none fully addresses the possibility

that substitution rates could differ markedly between any two

related lineages. As long as this possibility remains insufficiently

investigated and understood, reliance on molecular

estimates for the timing of diversification in mammals and

other groups seems unwarranted.

Figure 28.11. The topology implied by Kumar and Hedges’

(1998) molecular clock divergences (A) contradicts that of

published parsimony analyses of character data (B). Kumar and

Hedges’ (1998) molecular clock estimates state that a rodent

(Sciurognathi) split from humans at 112 ± 3.5 Mya and that

rabbits (Lagomorpha) split from humans 90.8 ± 2.0 Mya.

Rodents and rabbits have been shown to be more closely related

to each other than either taxon is to humans (Murphy et al.

2001). In order for the Kumar and Hedges’s (1998) dates to be

possible, rabbits would have to be more closely related to

humans than they are to rodents (A), a topology that contradicts

the tree generated from character data (B).

A B

Rodents Rabbits Humans Rodents Rabbits Humans

Building the Mammalian Sector of the Tree of Life 511

Conclusions

Discovering the mammalian section of the Tree of Life will

require an enormous push for collection of both morphological

and molecular data. We have outlined here a recommendation

that these data should be assembled into

supermatrices because this will create the strongest connection

between the resulting tree topology and the underlying

character data. We have also noted that advances in search

algorithms make it increasingly straightforward to analyze

thousands of taxa simultaneously, making a single supermatrix

for Mammalia (combining extinct and extant taxa), a

goal that is becoming increasingly within reach.

We have described how tree structure is extremely important

for reconstructing the time of origin of a clade; one

example of the many ways the mammalian sector of the Tree

of Life can be applied to other evolutionary questions. Still

other applications include understanding how characters

have transformed through time (optimization), information

that can even be used to reconstruct missing data (e.g., skin,

behavior) in fossils (e.g., O’Leary 2001). Investigations of

biogeographic area of origin and time of origin are also highly

dependant on a well-tested underlying tree.

The last decade in particular has witnessed an enormous

increase in the amount of molecular data available for phylogenetic

analysis, and this new work is greatly enhanced by

a sophisticated bioinformatics infrastructure, namely, the

publicly supported molecular sequence database known as

GenBank at the National Center for Biotechnology Information,

which makes molecular sequence data quickly and freely

available to investigators worldwide. This availability of raw

data for molecule-based phylogenetic analyses makes the construction

of molecular supermatrices relatively straightforward.

New raw data can be quickly compared and combined with

previously collected raw data for new phylogenetic analyses.

The fact that this database now supports multiple alignments

will make the synthesis even easier.

Morphological data, by contrast, currently are not supported

by an equivalent centralized database within which

raw observations from published morphological analyses are

organized and archived for future phylogenetic analysis. As

a result, systematists working with morphological data often

find themselves in the position of “recollecting” data

someone has amassed before. This is an unacceptable and

wasteful repetition of effort that is in part responsible for

restraining large-scale supermatrix analyses that combine

molecular and morphological matrices. We believe that the

databasing of morphological observations (homology statements)

must be improved and that this is one of the most

crucial modifications that must occur for a Tree of Life effort

to be successful, not just for Mammalia but for all species.

Our knowledge of extinct species, which far outnumber extant

species in the mammalian clade, also comes almost exclusively

from morphology. The full integration of molecular

and morphological data so critical to resolving problems in

mammal phylogeny will be most easily accomplished after

the development of an appropriate bioinformatics infrastructure

for archiving morphological data such as has been proposed

as MorphoBank (2003).

The recent explosion of published phylogenetic analyses

for many mammal clades includes contributions from

such historically disparate fields as histology, paleontology,

and molecular biology, challenges mammalian systematists

to absorb data collected outside their field of specialization.

Integration of these data will provide the greatest explanatory

power because it will cast phylogenetic analysis not as a

search for a subset of characters and taxa that will unlock

phylogenetic truth, but as an accretionary synthesis of detailed

comparative work across all phenotypic and genotypic

systems and in all taxa.

Acknowledgments

We thank J. Cracraft and M. Donoghue for inviting us to

contribute this article to the Tree of Life symposium at the

American Museum of Natural History. For helpful discussion or

comments on the manuscript, we thank K. de Queiroz, M.

Malia, Jr., and one anonymous reviewer. Figure 28.1 was

prepared by E. Heck with the assistance of L. Merrill; other

figures were prepared by L. Betti-Nash or the authors. This

work was supported by grants to M.A.O. (NSF DEB 9903964),

M.A. (NSF DEB-9629319; the research participation program of

the Oak Ridge Institute for Science and Education and the

Counterterrorism and Forensic Science Research Unit at the FBI

Academy), M.J.N. (NSF-DEB 9996172, NSF-DEB-0129031, the

Antorchas Foundation, and the Frick Laboratory Endowment),

J.G. (NSF-DEB 9985847), and J.M. (NSF-EAR-0120727 and

Chinese National Science Foundation grant 49928202).

Literature Cited

Adams, E. N., III. 1972. Consensus techniques and the

comparison of taxonomic trees. Syst. Zool. 21:390–397.

Allard, M. W., K. Miller, M. Wilson, K. L. Monson, and

B. Budowle. 2002. Characterization of the Caucasian

haplogroups present in the SWGDAM forensic mtDNA data

set for 1771 human control region sequences. J. Forens. Sci.

47:1215–1223.

Allard, M. W., M. M. Miyamoto, and R. L. Honeycutt. 1991.

Tests for rodent polyphyly. Nature 353:610–611.

Alroy, J. 1999. The fossil record of North American mammals:

evidence for a Paleocene evolutionary radiation. Syst. Biol.

48:107–118.

Archer, M., R. Arena, M. Bassarova, K. Black, J. Brammall,

B. Cooke, P. Creaser, K. Crosby, A. Gillespie, H. Godthelp,

et al.1999. The evolutionary history and diversity of

Australian mammals. Austral. Mammal. 21:1–45.

Archibald, J. D. 1996. Fossil evidence for a Late Cretaceous

origin of “hoofed” mammals. Science 272:1150–1153.

Archibald, J. D, A. O. Averianov, and E. G. Ekdale. 2001. Late

Cretaceous relatives of rabbits, rodents, and other extant

eutherian mammals. Nature 414:62–65.

512 The Relationships of Animals: Deuterostomes

Ayala, F. J. 1997. Vagaries of the molecular clock. Proc. Natl.

Acad. Sci. USA 94:7776–7783.

Ayala, F. J., A. Rzhetsky, and F. J. Ayala. 1998. Origin of the

metazoan phyla: molecular clocks confirm paleontological

estimates. Proc. Natl. Acad. Sci. USA 95:606–611.

Ax, P. 1987. The phylogenetic system: the systematization of

organisms on the basis of their phylogenies. Wiley, New

York.

Barrett, M., M. Donoghue, and E. Sober. 1991. Against

consensus. Syst. Zool. 40:486–493.

Baum, B. 1992. Combining trees as a way of combining data

sets for phylogenetic inference, and the desirability of

combining gene trees. Taxon 41:3–10.

Baum, B., and M. A. Ragan. 1993. Comment on Baum method for

combining phylogenetic trees—reply. Taxon 42:637–640.

Bremer, K. 1994. Branch support and tree stability. Cladistics

10:295–304.

Bull, J. J., J. P. Huelsenbeck, C. W. Cunningham, D. L. Swofford,

and P. J. Waddell. 1993. Partitioning and combining data in

phylogenetic analysis. Syst. Biol. 42:384–397.

Casiliano, M. L., and W. A. Clemens. 1979. Symmetrodonta.

Pp. 151–161 in Mesozoic mammals: the first two-thirds of

mammalian history (J. Lillegraven, Z. Kielan-Jaworowska,

and W. Clemens, eds.). University of California Press,

Berkeley.

Catzeflis, F. M. 1993. Mammalian phylogeny: morphology and

molecules. Trends Ecol. Evol. 8:340–341.

Cifelli, R. L. 1999. Tribosphenic mammal from the North

American Early Cretaceous. Nature 401:363–366.

Cifelli, R. L. 2000. Cretaceous mammals of Asia and North

America. Paleontol. Soc. Korea Spec. Publ. 4:49–84.

Cifelli, R. L. 2001. Early mammalian radiations. J. Paleontol.

75:1214–1226.

Cifelli, R. L., and C. De Muizon. 1997. Dentition and jaw of

Kokopellia juddi, a primitive marsupial or near-marsupial

from the medial Cretaceous of Utah. J. Mammal. Evol.

4:241–258.

Clemens, W. A., Jr. 1968. A mandible of Didelphodon vorax

(Marsupialia, Mammalia). Contrib. Sci. 133:1–11.

Cole, R., and R. Hariharan. 1996. An O(n log n) algorithm for

the maximum agreement subtree problem for binary trees.

Pp. 323–332 in Proceedings of the 7th annual ACM-SIAM

Symposium on Discrete Algorithms (SODA96).

Crompton, A. W. 1995. Masticatory function in nonmammalian

cynodonts and early mammals. Pp. 55–75 in Functional

morphology in vertebrate paleontology (J. J. Thomason,

ed.). Cambridge University Press, Cambridge.

Crompton, A. W., and F. A. Jenkins, Jr. 1973. Mammals from

reptiles: a review of mammalian origins. Annu. Rev. Earth

Planet. Sci. 1:131–155.

Crompton, A. W., and F. A. Jenkins, Jr. 1979. Origin of

mammals. Pp. 59–73 in Mesozoic mammals—the first twothirds

of mammalian history (J. A. Lillegraven, Z. Kielan-

Jaworowska, and W. A. Clemens, eds.). University of

California Press, Berkeley.

Cunningham, C. W. 1997. Is congruence between data

partitions a reliable predictor of phylogenetic accuracy?

Empirically testing an iterative procedure for choosing

among phylogenetic methods. Syst. Biol. 46:464–478.

Cutler, D. J. 2000. Estimating divergence times in the presence

of an overdispersed molecular clock. Mol. Biol. Evol.

17:1647–1660.

Dashzeveg, D., and Z. Kielan-Jaworowska. 1984. The lower jaw

of an aegialodontid mammal from the Early Cretaceous of

Mongolia. Zool. J. Linn. Soc. 82:217–227.

Dawson, M. R., and K. C. Beard. 1996. New late Paleocene

rodents (Mammalia) from Big Multi Quarry, Washakie

Basin, Wyoming. Palaeovertebrata 25:301–321.

Dawson, M. R., C. Li, and T. Qi. 1984. Eocene ctenodactyloid

rodents (Mammalia) of eastern central Asia. Spec. Publ.

Carnegie Mus. Nat. Hist. 9:138–150.

De Muizon, C., and R. L. Cifelli. 2000. The “condylarths”

(archaic Ungulata, Mammalia) from Early Palaeocene of

Tiupampa (Bolivia): implications on the origin of the South

American ungulates. Geodiversitas 22:47–150.

De Queiroz, K., and J. Gauthier. 1992. Phylogenetic taxonomy.

Annu. Rev. Ecol. Syst. 23:449–480.

DeSalle, R., and A. V. Z. Brower. 1997. Process partitions,

congruence, and the independence of characters: inferring

relationships among closely related Hawaiian Drosophila

from multiple gene regions. Syst. Biol. 46:751–764.

DeSalle, R., G. Giribet, and W. Wheeler (eds.). 2002. Techniques

in molecular systematics and evolution. Birkhдuser

Verlag, Boston.

Domning, D. P., C. E. Ray, and M. C. McKenna. 1986. Two new

Oligocene Desmostylians and a discussion of tethytherian

systematics. Smithson. Contrib. Paleobiol. 59:1–56.

Eizirik, E., W. J. Murphy, and S. J. O’Brien. 2001. Molecular

dating and biogeography of the early placental mammal

radiation. J. Hered. 92:212–219.

Farris, J. S. 1969. A successive approximations approach to

character weighting. Syst. Zool. 18:374–385.

Farris, J. S. 1983. The logical basis of phylogenetic analysis.

Pp.1–36 in Advances in cladistics (N. I. Platnick and V. A.

Funk, eds.), vol. 2. Columbia University Press, New York.

Felsenstein, J. 1985. Confidence limits on phylogenies: an

approach using the bootstrap. Evolution 39:783–791.

Fitch, W. M. 1977. The phyletic interpretation of macromolecular

sequence information: sample cases. Pp. 211–

248 in Major patterns of vertebrate evolution (M. K.

Hecht, P. C. Goody, and B. B. Hecht, eds.). Plenum Press,

New York

Flynn J. J., J. M. Parish, B. Rakotosamimanana, W. F. Simpson,

and A. R. Wyss. 1999. A middle Jurassic mammals from

Madagascar. Nature 401:57–60.

Foote, M., J. P. Hunter, C. M. Janis, and J. J. Sepkoski. 1999.

Evolutionary and preservational constraints on origins of

biologic groups: divergence times of eutherian mammals.

Science 283:1310–1314.

Fox, R. C., G. P. Youzwyshyn, and D. W. Krause. 1992. Post-

Jurassic mammal-like reptile from the Palaeocene. Nature

358:233–235.

Gatesy, J., C. Matthee, R. DeSalle, and C. Hayashi. 2002.

Resolution of a supertree/supermatrix paradox. Syst. Biol.

51:652–664.

Gatesy, J., M. Milinkovitch, V. Waddell, and M. Stanhope.

1999a. Stability of cladistic relationships between Cetacea

and higher level artiodactyl taxa. Syst. Biol. 48:6–20.

Gatesy, J., P. O’Grady, and R. H. Baker. 1999b. Corroboration

among data sets in simultaneous analysis: hidden support

Building the Mammalian Sector of the Tree of Life 513

for phylogenetic relationships among higher level artiodactyl

taxa. Cladistics 15:271–314.

Gatesy, J., and M. A. O’Leary. 2001. Deciphering whale origins

with molecules and fossils. Trends Ecol. Evol. 16:562–570.

Gauthier, J., A. G. Kluge, and T. Rowe. 1988. Amniote phylogeny

and the importance of fossils. Cladistics 4:105–209.

Geisler, J. H. 2001. New morphological evidence for the

phylogeny of Artiodactyla, Cetacea, and Mesonychidae. Am.

Mus. Nov. 3344:1–53.

GenBank. 2003. Molecular data archive. Available: www.ncbi.

nlm.nih.gov. Last accessed 25 December 2003.

Goloboff, P. 1994. NONA: a tree search program, vers. 2

(program and documentation). Available: www.cladistics.

com. Last accessed 8 December 2003.

Goloboff, P., and J. S. Farris. 2001. Methods for quick consensus

estimation. Cladistics 17:s26–s34.

Goloboff, P., J. S. Farris, and K. Nixon. 1999. T.N.T.: tree

analysis using new technology. Available: http://www.

cladistics.com. Last accessed 8 December 2003.

Goodman, M., A. E. Romero-Herrara, H. Dene, J. Czeluzniak,

and R. E. Tashian. 1982. Amino acid sequence evidence on

the phylogeny of primates and other eutherians. Pp. 115–

191 in Macromolecular sequences systematic and evolutionary

biology (M. Goodman, ed.). Plenum Press, New York.

Graur, D. 1993. Molecular phylogeny and the higher classi-fication

of eutherian mammals. Trends Ecol. Evol. 8:141–147.

Graur, D., L. Duret, and M. Gouy. 1996. Phylogenetic position

of the order Lagomorpha (rabbits, hares and allies). Nature

379:333–335.

Graur, D., W. A. Hide, and W. H. Li. 1991. Is the guinea-pig a

rodent. Nature 351:649–652.

Gregory, W. K. 1947. The monotremes and the palimpsest

theory. Bull. Am. Mus. Nat. Hist. 88:1–52.

Halanych, K. M. 1998. Lagomorphs misplaced by more

characters and fewer taxa. Syst. Biol. 47:138–146.

Hasegawa, M., Y. Cao, J. Adachi, and T.-A. Yano. 1992. Rodent

polyphyly. Nature 355:595.

Hedges, S. B., and L. R. Maxson. 1996. Re: molecules and

morphology in amniote phylogeny. Mol. Phylogenet. Evol.

6:312–314.

Hedges, S. B., P. H. Parker, C. G. Sibley, and S. Kumar. 1996.

Continental breakup and the ordinal diversification of birds

and mammals. Nature 381:226–229.

Hedges, S. B., and L. L. Poling. 1999. A molecular phylogeny of

reptiles. Science 283:998–1001.

Hopson, J. A. 1994. Synapsid evolution and the radiation of

non-eutherian mammals. Pp. 190–219 in Major features of

vertebrate evolution (R. S. Spencer, ed.). Paleontological

Society, Knoxville, TN.

Hu, Y. M., Y. Q. Wang, Z. X. Luo, and C. K. Li. 1997. A new

symmetrodont mammal from China and its implications for

mammalian evolution. Nature 390:137–142.

Janies, D., and W. Wheeler. 2001. Efficiency of parallel direct

optimization. Cladistics 17:S71–S82.

Janies, D., and W. Wheeler. 2002. Theory and practice of

parallel direct optimization. Pp. 115–123 in Molecular

systematics and evolution (R. DeSalle, W. Wheeler, and

G. Giribet, eds.). Birkhauser Verlag, Boston.

Janke, A., G. Feldmaier-Fuchs, W. K. Thomas, A. von-Haeseler,

and S. Pддbo. 1994. The marsupial mitochondrial genome

and the evolution of placental mammals. Genetics 137:243–

256.

Janke, A., O. Magnell, G. Wieczorek, M. Westerman, and U.

Arnason. 2002. Phylogenetic analysis of 18S rRNA and the

mitochondrial genomes of the wombat, Vombatus ursinus,

and the spiny anteater, Tachyglossus aculeatus: increased

support for the Marsupionta hypothesis. J. Mol. Evol.

54:71–80.

Janke, A., X. Xu, and U. Arnason. 1997. The complete mitochondrial

genome of the wallaroo (Macropus robustus) and

the phylogenetic relationship among Monotremata,

Marsupialia, and Eutheria. Proc. Natl. Acad. Sci. USA

94:1276–1281.

Jefferies, R. P. S. 1979. The origin of chordates—a methodological

essay. Pp. 443–477 in The origin of major invertebrate

groups (M. R. House, ed.). Systematics Association spec.

vol. 12. Academic Press, New York.

Jenkins, F., Jr., and D. W. Krause. 1983. Adaptations for

climbing in North American multituberculates (Mammalia).

Science 220:712–715.

Ji, Q., Z. Luo, and S.-A. Ji. 1999. A Chinese triconodont

mammal and mosaic evolution of the mammalian skeleton.

Nature 398:326–330.

Ji, Q., Z.-X. Luo, C.-X. Yuan, J. R. Wible, and J.-P. Zhang, and J.

A. Georgi. 2002. The earliest known eutherian mammal.

Nature 416:816–822.

Kearney, M. 2002. Fragmentary taxa, missing data, and

ambiguity: mistaken assumptions and conclusions. Syst.

Biol. 51:369–381.

Kermack, D. M., K. A. Kermack, and F. Mussett. 1968. The

Welsh pantothere Kuehneotherium praecursoris. J. Linn. Soc.

Lond. Zool. 47:407–422.

Kermack, K. A., and F. Mussett. 1958. The jaw articulation of

the Docodonta and the classification of Mesozoic mammals.

Proc. R. Soc. Lond. B. 148:535–554.

Kermack, K. A., F. Mussett, and H. W. Rigney. 1973. The lower

jaw of Morganucodon. Zool. J. Linn. Soc. 53:87–175.

Kielan-Jaworowska, Z., R. L. Cifelli, and Z. Luo. 1998. Alleged

Cretaceous plcental from down under. Lethaia 31:267–268.

Kitching I. J., P. L. Forey, C. J. Humphries, and D. M. Williams.

1998. Cladistics (2nd ed.): the theory and practice of

parsimony analysis. Systematics Association publ. no. 11.

Oxford Science Publications, New York.

Kluge, A. G. 1989. A concern for evidence and a phylogenetic

hypothesis of relationships among Epicrates (Boidae,

Serpentes). Syst. Zool. 38:7–25.

Korth, W. W. 1984. Earliest Tertiary evolution and radiation of

rodents in North America. Bull. Carnegie Mus. Nat. Hist.

24:1–71.

Krebs, B. 1971. Evolution of the mandible and lower dentition

in dryolestids (Pantotheria, Mammalia). Pp. 89–102 in Early

mammals (D. M. Kermack and K. A. Kermack, eds.).

Zoological Journal of the Linnaean Society, vol. 50, suppl. 1.

Kьhne, W. G. 1973. The systematic position of monotremes

reconsidered. Z. Morphol. Tiere 75:59–64.

Kumar, S., and S. B. Hedges. 1998. A molecular timescale for

vertebrate evolution. Nature 392:917–920.

Langer, P. 2001. Evidence from the digestive tract on phylogenetic

relationships in ungulates and whales. J. Zool. Syst.

Evol. Res. 39:77–90.

514 The Relationships of Animals: Deuterostomes

Lee, M. S. Y. 1999. Molecular clock calibrations and metazoan

divergence dates. J. Mol. Evol. 49:385–391.

Li, C. K., R. W. Wilson, M. R. Dawson, and L. Krishtalka. 1987.

The origin of rodents and lagomorphs. Pp. 91–108 in

Current mammalogy, vol. 1 (H. H. Genoways, ed.). Plenum

Press, New York.

Li, W.-H. 1997. Molecular evolution. Sinauer, Sunderland, MA.

Li, W.-H., and D. Graur. 1991. Fundamentals of molecular

evolution. Sinauer, Sunderland, MA.

Li, W.-H., W. A. Hide, and D. Graur. 1992. Origin of rodents

and guinea-pigs. Nature 359:277–278.

Lipscomb, D. L., J. S. Farris, M. Kдllersjц, and A. Tehler. 1998.

Support, ribosomal sequences and the phylogeny of the

Eukaryotes. Cladistics 14:303–338.

Liu, F. R., M. M. Miyamoto, N. P. Freire, P. Q. Ong, M. R.

Tennent, T. S. Young, and K. F. Gugel. 2001. Molecular and

morphological supertrees for eutherian (placental) mammals.

Science 291:1786–1789.

Luckett, W. P., and J.-L. Hartenberger. 1993. Monophyly or

polyphyly of the order Rodentia: possible conflict between

morphological and molecular interpretations. J. Mammal.

Evol. 1:127–147.

Luckett, W. P., and N. Hong. 1998. Phylogenetic relationships

between the orders Artiodactyla and Cetacea: a combined

assessment of morphological and molecular evidence. J.

Mammal. Evol. 5:127–182.

Luo, Z.-X., R. L. Cifelli, and Z. Kielan-Jaworowska. 2001a. Dual

origin of tribosphenic mammals. Nature 409:53–57.

Luo, Z.-X., A. W. Crompton, and A.-L. Sun. 2001b. A new

mammaliaform from the Early Jurassic and evolution of

mammalian characteristics. Science 292:1535–1540.

Luo, Z.-X., Z. Kielan-Jaworowska, and R. L. Cifelli. 2002. In

quest for a phylogeny of Mesozoic mammals. Acta Palaeontol.

Pol. 47:1–78.

Madsen, O., M. Scally, C. J. Douady, D. J. Kao, R. W. DeBry,

R. Adkins, H. M. Amrine, M. J. Stanhope, W. W. de Jong,

and M. S. Springer. 2001. Parallel adaptive radiations in two

major clades of placental mammals. Nature 409:610–614.

Malia, M. J., Jr., D. L. Lipscomb, and M. W. Allard. 2003. The

misleading effects of composite taxa in supermatrices. Mol.

Phylogenet. Evol. 27:522–527.

Marshall, C. R. 1990. The fossil record and estimating divergence

times between lineages: maximum divergence times

and the importance of reliable phylogenies. J. Mol. Evol.

30:400–408.

Matthee, C. A., J. D. Burzlaff, J. F. Taylor, and S. K. Davis. 2001.

Mining the mammalian genome for artiodactyl systematics.

Syst. Biol. 50:367–390.

McDowell, S. B. 1958. The Greater Antillean insectivores. Bull.

Am. Mus. Nat. Hist. 115:117–214.

McKenna, M. C. 1982. Lagomorpha interrelationships. Geobios.

Mem. Spec. 6:213–223.

McKenna, M. C., and S. K. Bell. 1997. Classification of mammals

above the species level. Columbia University Press,

New York.

Meng, J., Y.-M. Hu, and C. Li. 2003. The osteology of

Rhombomylus (Mammalia, Glires) and its implication for

Glires systematics and evolution. Bull. Am. Mus. Nat. Hist.

275:1–247.

Meng, J., and A. Wyss. 1997. Multituberculate and other

mammal hair recovered from Palaeogene excreta. Nature

385:712–714.

Meng, J., and A. R. Wyss. 2001. The Morphology of

Tribosphenomys (Rodentiaformes, Mammalia): phylogenetic

implications for basal Glires. J. Mammal. Evol. 8:1–71.

Messenger, S. L., and J. A. McGuire. 1998. Morphology,

molecules, and the phylogenetics of cetaceans. Syst. Biol.

47:90–124.

Miyamoto, M. M., and W. M. Fitch. 1995. Testing species

phylogenies and phylogenetic methods with congruence.

Syst. Biol. 44:64–76.

MorphoBank. 2003. Available: www.morphobank.net.

Murphy, W. J., E. Eizirik, W. E. Johnson, Y.-P. Zhang, O. A.

Ryder, and S. J. O’Brien. 2001. Molecular phylogenetics and

the origins of placental mammals. Nature 409:614–618.

Murphy, W. J., E. Eizirik, S. J. O’Brien, O. Madsen, M. Scally,

C. J. Douady, E. Teeling, O. A. Ryder, M. J. Stanhope,

W. W. de Jong, and M. S. Springer. 2002. Resolution of the

early placental mammal radiation using Bayesian

phylogenetics. Science 294:2348–2351.

Musser, A. M., and M. Archer. 1998. New information about the

skull and dentary of the Miocene platypus Obdurodon

dicksoni, and a discussion of ornithorhynchid relationships.

Philos. Trans. R. Soc. Lond. Biol. Sci. B 353:1063–1079.

Naylor, G. J. P., and D. Adams. 2001. Are the fossil data really

at odds with the molecular data? Morphological evidence

for Cetartiodactyla phylogeny reexamined. Syst. Biol.

50:444–453.

Naylor, G. J. P., and W. M. Brown. 1998. Structural biology and

phylogenetic estimation. Nature 388:527–528.

Nikaido, M., F. Matsuno, H. Hamilton, R. L. Brownell, Jr.,

Y. Cao, W. Ding, Z. Zuoyan, A. M. Shedlock, R. E. Fordyce,

M. Hasegawa, and N. Okada. 2001. Retroposon analysis of

major cetacean lineages: the monophyly of toothed whales

and the paraphyly of river dolphins. Proc. Natl. Acad. Sci.

USA 98:7384–7389.

Nikaido, M., A. P. Rooney, and N. Okada. 1999. Phylogenetic

relationships among cetartiodactyls based on insertions of

short and long interspersed elements: hippopotamuses are

the closest extant relatives of whales. Proc. Natl. Acad. Sci.

USA 96:10261–10266.

Nixon, K. C. 1999. The parsimony ratchet, a new method for

rapid parsimony analysis. Cladistics 15:407–414.

Nixon, K. C. 2002. WinClada, ver. 1.00.08. K. C. Nixon, Ithaca,

NY. Available: http://www.cladistics.com. Last accessed 25

December 2003.

Nixon, K. C., and J. M. Carpenter. 1996. On simultaneous

analysis. Cladistics 12:221–241.

Nixon, K. C., and J. M. Carpenter. 2000. On the other “phylogenetic

systematics.” Cladistics 16:298–318.

Norell, M. A. 1992. Taxic origin and temporal diversity: the

effect of phylogeny. Pp. 89–118 in Extinction and phylogeny

(M. J. Novacek and Q. D. Wheeler, eds.). Columbia

University Press, New York.

Novacek, M. J. 1982. Information for molecular studies from

anatomical and fossil evidence on higher eutherian

phylogeny. Pp. 3–41 in Macromolecular sequences in

systematic and evolutionary biology (M. Goodman, ed.).

Plenum Press, New York.

Novacek, M. J. 1992a. Fossils, topologies, missing data, and the

Building the Mammalian Sector of the Tree of Life 515

higher level phylogeny of eutherian mammals. Syst. Biol.

41:58–73.

Novacek, M. J. 1992b. Mammalian phylogeny: shaking the tree.

Nature 356:121–125.

Novacek, M. J. 1993. Reflections on higher mammalian

phylogenetics. J. Mammal. Evol. 1:3–30.

Novacek, M. J. 1999. 100 million years of land vertebrate

evolution: the Cretaceous-Early Tertiary transition. Ann.

Mo. Bot. Gard. 86:230–258.

Novacek, M. J., G. W. Rougier, J. R. Wible, M. C. McKenna,

D. Dashzeveg, and I. Horovitz. 1997. Epipubic bones in

eutherian mammals from the Late Cretaceous of Mongolia.

Nature 389:483–486.

Novacek, M. J., and A. R. Wyss. 1987. Selected features of the

desmostylian skeleton and their phylogenetic implications.

Am. Mus. Nov. 2870:1–8.

Nowak, R. M. 1999. Walker’s mammals of the world. 6th ed.

Johns Hopkins University Press, Baltimore, MD.

O’Leary, M. A. 2001. The phylogenetic position of cetaceans:

further combined data analyses, comparisons with the

stratitgraphic record and a discussion of character optimization.

Am. Zool. 41:487–506.

O’Leary, M. A., J. Gatesy, and M. J. Novacek. 2003. Are the

dental data really at odds with the molecular data? Morphological

evidence for whale phylogeny (re)reexamined. Syst.

Biol. 52:853–864.

O’Leary, M. A., and J. H. Geisler. 1999. The position of Cetacea

within Mammalia: phylogenetic analysis of morphological

data from extinct and extant taxa. Syst. Biol. 48:455–490.

O’Leary, M. A., and M. D. Uhen. 1999. The time of origin of

whales and the role of behavioral changes in the terrestrialaquatic

transition. Paleobiology 25:534–556.

Olson, E. C. 1944. Origin of mammals based upon cranial

morphology of the therapsid suborders. Geol. Soc. Am.

Spec. Pap. 55:1–136.

Olson, E. C. 1959. The evolution of mammalian characters.

Evolution 13:344–353.

Padian, K., and J. M. V. Rayner. 1993. The wings of pterosaurs.

Am. J. Sci. 293A:91–166.

Pascual, R., F. J. Goin, D. W. Krause, E. Ortiz-Jaureguizar, and

A. A. Carlini. 1999. The first gnathic remains of Sudamerica:

Implications for gondwanathere relationships. J. Vert.

Paleontol. 19:373–382.

Pisani, D., and M. Wilkinson. 2002. Matrix representation with

parsimony, taxonomic congruence, and total evidence. Syst.

Biol. 51:151–155.

Prendini, L. 2001. Species or supraspecific taxa as terminals in

cladistic analysis? Groundplans versus exemplars revisited.

Syst. Biol. 50:290–300.

Purvis, A. 1995. A modification to Baum and Ragan’s method

for combining phylogenetic trees. Syst. Biol. 44:251–255.

Ragan, M. A. 1992. A phylogenetic inference based on matrix

representation of trees. Mol. Phylogenet. Evol. 1:53–58.

Rauhut, O. W. M., T. Martin, E. Ortiz-Jaureguizar, and P.

Puerta. 2002. A Jurassic mammal from South America.

Nature 416:165–168.

Rice, K. A., M. J. Donoghue, and R. G. Olmstead. 1997.

Analyzing large data sets: rbcL 500 revisited. Syst. Biol.

46:554–563.

Rich, T. H., T. F. Flannery, P. Trusler, A. L. Kool, N. van

Klaveren, and P. Vickers-Rich. 2001a. Early Cretaceous

mammals from Flat Rocks, Victoria, Australia. Rec. Queen

Victoria Mus. 106:1–30.

Rich, T. H., P. Vickers-Rich, A. Constantine, T. F. Flannery,

L. Kool, and N. van Klaveren. 1997. A tribosphenic

mammal from the Mesozoic of Australia. Science 278:1438–

1442.

Rich, T. H., P. Vickers-Rich, P. Trusler, T. F. Flannery, R.

Ciffelli, A. Constantine, L. Kool, and N. van Klaveren.

2001b. Corroboration of the Garden of Eden hypothesis.

Pp. 324–332 in Faunal and floral migrations and evolution

in SE Asia-Australia (I. Metcalfe, J. M. B. Smith, M.

Morwood, I. Davidson, and K. Hewison, eds.). A. A.

Balkema, Lisse, the Netherlands.

Riggs, E. S. 1935. A skeleton of Astrapotherium [Argentina].

Field Mus. Nat. Hist. Geol. Ser. 6. 13:167–176.

Rodrнguez-Trelles, F., R. Tarrнo, and F. J. Ayala. 2001. Erratic

overdispersion of three molecular clocks: GPDH, SOD, and

XDH. Proc. Natl. Acad. Sci. USA. 98:11405–11410.

Rodrнguez-Trelles, F., R. Tarrнo, and F. J. Ayala. 2002. A

methodological bias toward overestimation of molecular

evolutionary time scales. Proc. Natl. Acad. Sci. USA

99(12):8112–8115.

Ronquist, F. 1996. Matrix representation of trees, redundancy

and weighting. Syst. Biol. 45:247–253.

Rose, K. D. 1987. Climbing adaptations in the early Eocene

mammal Chriacus and the origin of Artiodactyla. Science

236:314–316.

Rougier, G. W. 1993. Vincelestes neuquenianus Bonaparte

(Mammalia, Theria) un primitivo mamifero del cretacio

inferior de la cuenca Neuquina. 1–3. Ph.D. thesis,

Universidad de Buenos Aires.

Rougier, G. W., J. R. Wible, and J. A. Hopson. 1996. Basicranial

anatomy of Priacodon fruitaensis (Triconodontidae, Mammalia)

from the Late Jurassic of Colorado, and a reappraisal of

mammaliaform interrelationships. Am. Mus. Nov. 3183:1–38.

Rougier, G. W., J. R. Wible, and M. J. Novacek. 1998. Implications

of Deltatheridium specimens for early marsupial

history. Nature 396:459–463.

Rowe, T. 1988. Definition, diagnosis, and origin of Mammalia.

J. Vert. Paleontol. 8:241–264.

Salamin, N., T. R. Hodkinson, and V. Savolainen. 2002.

Building supertrees: an empirical assessment using the grass

family (Poaceae). Syst. Biol. 51:136–150.

Sanderson, M. J. 1997. A nonparametric approach to estimating

divergence times in the absence of rate constancy. Mol. Biol.

Evol. 14:1218–1231.

Sanderson, M. J., A. Purvis, and C. Henze. 1998. Phylogenetic

supertrees: assembling the trees of life. Trends Ecol. Evol.

13:105–109.

Sarich, V. M., and A. C. Wilson. 1967. Immunological time

scale for hominid evolution. Science 158:1200–1203.

Scott, W. B. 1910. Litopterna of the Santa Cruz Beds. Rep.

Princeton Univ. Exped. Patagonia, 1896–1899. 7:1–156.

Shoshani, J., and M. C. McKenna. 1998. Higher taxonomic

relationships among extant mammals based on morphology,

with selected comparisons of results from molecular data.

Mol. Phylogenet. Evol. 9:572–584.

Simpson, G. G. 1936. Studies of the earliest mammalian

dentitions. Dental Cosmos 6:1–24.

516 The Relationships of Animals: Deuterostomes

Simpson, G. G. 1960. Diagnosis of the Classes Reptilia and

Mammalia. Evolution 14:388–392.

Simpson, G. G. 1967. The beginning of the age of mammals in

South America, Part 2. Bull. Am. Mus. Nat. Hist. 137:1–260.

Sinclair, W. J. 1906. Marsupials of the Santa Cruz Beds. Rep.

Princeton Univ. Exped. Patagonia. 1896–1899. 4:333–

359.

Smith, A. B., and K. J. Peterson. 2002. Dating the time of origin

of major clades: molecular clocks and the fossil record.

Annu. Rev. Earth Planet. Sci. 30:65–88.

Springer, M. S. 1997. Molecular clocks and the timing of the

placental and marsupial radiations in relation to the Cretaceous-

Tertiary boundary. J. Mammal. Evol. 4:285–302.

Springer, M. S., W. J. Murphy, E. Eizirik, and S. J. O’Brien.

2003. Placental mammal diversification and the Cretaceous-

Tertiary boundary. Proc. Natl. Acad. Sci. USA 100:1056–

1061.

Sullivan, J., and D. L. Swofford. 1997. Are guinea pigs rodents?

The importance of adequate models in molecular phylogenetics.

J. Mammal. Evol. 4:77–86.

Swofford, D. L. 2000. PAUP*: phylogenetic analysis using

parsimony (and other methods), vers. 4 beta. Sinauer,

Sunderland, MA.

Tajima, F. 1993. Simple methods for testing the molecular clock

hypothesis. Genetics 135:599–607.

Thewissen, J. G. M., E. M. Williams, and S. T. Hussain. 2001.

Skeletons of terrestrial cetaceans and the relationships of

whales to artiodactyls. Science 413:277–281.

Thorne, J. L., H. Kishino, and I. S. Painter. 1998. Estimating the

rate of evolution of the rate of molecular evolution. Mol.

Biol. Evol. 15:1647–1657.

Unwin, D. M., and N. N. Bakhurina. 1994. Sordes pillosus and the

nature of the pterosaur flight apparatus. Nature 371:62–64.

Vaughan, T. A. 1986. Mammalogy. Saunders, Forth Worth, TX.

Waddell, P. J., H. Kishino, and R. Ota. 2001. A phylogenetic

foundation for comparative mammalian genomics. Genome

Informatics 12:141–154.

Waddell, P. J., N. Okada, and M. Hasegawa. 1999. Towards

resolving the interordinal relationships of placental

mammals. Syst. Biol. 48:1–6.

Wang, Y., Y. Hu, J. Meng, and C. Li. 2001. An ossified Meckel’s

cartilage in two Cretaceous mammals and origin of the

mammalian middle ear. Science 294:357–361.

Wheeler, W. C., and D. Gladstein. 2000. POY: the optimization

of alignment characters (program and documentation).

American Museum of Natural History, New York. Available:

www.cladistics.org. Last accessed 25 December 2003.

Wible, J. R., M. J. Novacek, and G. W. Rougier. 2004. New data

on the skull and dentition in the Mongolian Late Cretaceous

mammal Zalambdalestes. Bull. Am. Mus. Nat. Hist. 281:1–

144.

Wible, J. R., G. W. Rougier, M. J. Novacek, M. C. Mckenna, and

D. D. Dashzeveg. 1995. A mammalian petrosal from the

Early Cretaceous of Mongolia: implications for the evolution

of the ear region and mammaliamorph relationships. Am.

Mus. Nov. 3149:1–19.

Wilkinson, M., J. Thorley, D. Littlewood, and R. Bray. 2001.

Towards a phylogenetic supertree of Platyhelminthes?

Pp. 292–301 in Interrelationships of the Platyhelminthes

(D. Littlewood and R. Bray, eds.). Chapman-Hall, London.

Wilson, D. E., and D. M. Reeder. In press. Mammal species of

the world. A taxonomic and geographic reference. 3rd ed.

Smithsonian Institution Press, Washington, DC.

Wood, A. E. 1962. The early Tertiary rodents of the family

Paramyidae. Trans. Am. Philos. Soc. Phil. 52:1–261.

Wu, C.-I., and W.-H. Li. 1985. Evidence for higher rates of

nucleotide substitution in rodents than in man. Proc. Natl.

Acad. Sci. USA 82:1741–1745.

Zuckerkandl, E., and L. Pauling. 1962. Molecular disease,

evolution and genic heterogeneity. Pp. 189–225 in Horizons

in biochemistry (M. Kasha and B. Pullman, eds.). Academic

Press, New York.

Zuckerkandl, E., and L. Pauling. 1965. Evolutionary divergence

and convergence in proteins. Pp. 97–166 in Evolving genes

and proteins (V. Bryson and H. J. Vogel, eds.). Academic

Press, New York.

Навигация

• Главная
• Книги
• Статьи
• Обратная связь
• Поиск