ORIGINAL ARTICLE
doi:10.1111/j.1558-5646.2007.00077.x
LINEAGES WITH LONG DURATIONS ARE
OLD AND MORPHOLOGICALLY AVERAGE:
AN ANALYSIS USING MULTIPLE DATASETS
Lee Hsiang Liow1,2,3
1 Committee
2 E-mail:
on Evolutionary Biology, University of Chicago, 5734 S. Ellis Avenue, Chicago, Illinois 60637
l.h.liow@bio.uio.no
Received July 31, 2006
Accepted December 6, 2006
Lineage persistence is as central to biology as evolutionary change. Important questions regarding persistence include: why do some
lineages outlive their relatives, neither becoming extinct nor evolving into separate lineages? Do these long-duration lineages have
distinctive ecological or morphological traits that correlate with their geologic durations and potentially aid their survival? In this
paper, I test the hypothesis that lineages (species and higher taxa) with longer geologic durations have morphologies that are more
average than expected by chance alone. I evaluate this hypothesis for both individual lineages with longer durations and groups of
lineages with longer durations, using more than 60 published datasets of animals with adequate fossil records. Analyses presented
here show that groups of lineages with longer durations fall empirically into one of three theoretically possible scenarios, namely:
(1) the morphology of groups of longer duration lineages is closer to the grand average of their inclusive group, that is, their relative
morphological distance is smaller than expected by chance alone, when compared with rarified samples of their shorter duration
relatives (a negative group morpho-duration distribution); (2) the relative morphological distance of groups of longer duration
lineages is no different from rarified samples of their shorter duration relatives (a null group morpho-duration distribution); and
(3) the relative morphological distance of groups of longer duration lineages is greater than expected when compared with rarified
samples of their shorter duration relatives (a positive group morpho-duration distribution). Datasets exhibiting negative group
morpho-duration distributions predominate. However, lineages with higher ranks in the Linnean hierarchy demonstrate positive
morpho-duration distributions more frequently. The relative morphological distance of individual longer duration lineages is no
different from that of rarified samples of their shorter duration relatives (a null individual morpho-duration distribution) for the
majority of datasets studied. Contrary to the common idea that very persistent lineages are special or unique in some significant
way, both the results from analyses of long-duration lineages as groups and individuals show that they are morphologically
average. Persistent lineages often arise early in a group’s history, even though there is no prior expectation for this tendency in
datasets of extinct groups. The implications of these results for diversification histories and niche preemption are discussed.
KEY WORDS:
Average morphology, fossil record, lineage duration, prolonged stasis, survivorship.
Variation is rampant in the biological world. This variation
is present in lineage (species and higher taxa) durations: it
has been repeatedly observed that lineage durations of natu3 Current address: Centre for Ecological and Evolutionary Synthesis
(CEES), Department of Biology, P.O. Box 1066, Blindern, 0316 Oslo,
Norway. E-mail: l.h.liow@bio.uio.no
C
885
ral groups resemble hollow curves (Simpson 1944, 1953; Stanley 1979; Levinton and Ginzburg 1984). In any given clade or
paraclade (sensu Raup 1985, p. 44, e.g., monophyletic birds
or paraphyletic nonavian dinosaurs), most lineages have shorter
durations and a few have longer durations. This observation
leads naturally to the question: do lineages that outlive their
relatives without becoming extinct or evolving into separate
C 2007 The Society for the Study of Evolution.
2007 The Author(s). Journal compilation Evolution 61-4: 885–901
LEE HSIANG LIOW
lineages have distinctive properties that could aid their prolonged survival?
Factors that affect lineage durations or survivorship may be
categorized as extrinsic (environmental) or intrinsic (biological),
although the two categories often overlap. Biological characteristics may operate on different lineages in different ways during time
intervals of varying environmental conditions (e.g., mass or background extinction regimes), resulting in different lineage durations or survivorship (Jablonski 1986a, 1994; Jablonski and Raup
1995). Broad geographic ranges (Jackson 1974; Boucot 1975;
Chen et al. 2005; Jablonski 2005, Liow 2007), the possession
of planktotrophic larvae (Hansen 1978; Jablonski 1986b; Jeffery
and Emlet 2003) as well as deeper and wider depth distribution
of marine organisms (Buzas and Culver 1984; Oji 1996), high
site occupancy (fossil locality coverage; Jernvall and Fortelius
2004), generalist feedings strategies (Baumiller 1993), broader
niche breadth (Kammer et al. 1997, 1998), and greater ecological
tolerances (Jackson 1974; Schopf 1994) are some biological characteristics that promote lineage persistence and/or damp lineage
differentiation.
Morphology affects the functioning and performance of organisms (Koehl 1996) and reflects aspects of physiology and ecology (Wainright and Reilly 1994). Therefore, morphology could
serve as a proxy for ecology, which in turn affects not only the
longevity of individual organisms but also the survivorship of lineages. Yet, little is known about the distribution of morphology
in relation to lineage duration, much less the processes that contribute to it. Previous studies have focused on the relationship between morphological complexity and duration with mixed results:
the correlation of complexity with durations is dependent on the
clades analyzed and complexity metrics used (Flessa et al. 1975;
Anstey 1978; Ward and Signor 1983; Boyajian and Lutz 1992).
Lineages with longer durations might be morphologically
distant from the average morphology of their broader systematic grouping because being different may confer a competitive
advantage (a positive morpho-duration distribution; Fig. 1A). If
several related (and hence ecologically and/or morphologically
similar) lineages overlap in space and time, then those lineages
that are able to substantially differentiate themselves ecologically
may out-compete and survive beyond the average life spans of
their relatives. Conversely, lineages with longer durations might
also be morphologically closer to the average morphology of their
broader systematic grouping than expected because generalists
may be able to survive and persist through a greater range of environmental changes (Simpson 1944; Liow 2006; Fig. 1B, a negative morpho-duration distribution). Finally, lineages with shorter
or longer durations may not have significantly different distributions of morphology, indicating that factors unassociated with
morphology may be operating more strongly in influencing survivorship (Fig. 1C, a null morpho-duration distribution). Recent
886
EVOLUTION APRIL 2007
studies (Liow 2004, 2006) have concluded that crinoid and ostracode lineages in general show null morpho-duration distributions
(Fig. 1C). Similarly, morphological distances from the centroid of
morphospace of ammonoid survivors across the Permian-Triassic
extinction are not significantly different from those of victims
(McGowan, in press). These results are contrary to the long-held
idea that geologically persistent or “living fossil” taxa have special or distinctive properties that enable them to remain unchanged
whereas their relatives experience speciation and extinction (e.g.,
see Wills 2001).
Previous attempts to investigate the relationship between
morphological distribution and lineage duration used only crinoid
genera and families (Liow 2004) and trachyleberidid ostracode
genera (Liow 2006). Hence taxonomic coverage was very limited. To test the validity of either of the three morpho-duration
distributions (Fig. 1) as a general evolutionary pattern, this paper uses multiple datasets to move toward a more representative
sample of phylogenetically independent para(clades) across more
branches of the metazoan tree of life. These datasets span a wide
range of body plans, ecologies, geologic ranges, and lineage ages.
Datasets where species are the units of study were also included in
the current analysis, in contrast with previous studies where only
genus and family data were available.
A methodological limitation of previous attempts to investigate the relationship between morphological distribution and lineage duration (Liow 2004, 2006) is that a continuous variable
(lineage duration) was arbitrarily divided into discrete categories
(shorter and longer). To overcome this limitation, I developed
a sequential rarefaction method to determine whether there are
tendencies for (para)clades to show one of the three continuous
morpho-duration distributions illustrated in Figure 1. In addition,
the hypothesis that lineages with longer durations have morphologies that are closer than expected to the grand average of their
inclusive group (i.e., a shorter relative morphological distance),
is tested both when these lineages are considered collectively as
a group (group analyses) and when they are considered individually (single analyses). This is because both individual lineages
and groups of lineages with long durations could be unique or
distinctive in comparison with their shorter duration counterparts.
The datasets used in this study differ in the numbers of
lineages, numbers and types of morphological characters represented, operational taxonomic units, intrinsic probabilities of
geologic preservation, geologic age ranges, and lineage duration
distributions. All of these factors may bias morpho-duration distributions but have never been analyzed within a comparative analytical framework. I also test whether lineages with longer durations
tend to arise significantly earlier in the history of their inclusive
group. Biases and possible drawbacks in using multiple published
datasets are presented for transparency. I end by discussing numerous macroevolutionary implications of the results.
Probabilistic relative morphological distance of lineages
MORPHOLOGY AND LINEAGE DURATIONS
A.
Positive morphoduration distribution
B.
Negative morphoduration distribution
C.
Null morpho-duration
distribution
Lineage duration
The three general relationships between the probabilistic relative morphological distances of lineages (from a grand morphological average) versus their lineage durations (i.e., morphoduration distributions). Panel A shows a positive morpho-duration
distribution, panel B shows a negative morpho-duration distribution, and panel C shows a null morpho-duration distribution.
Figure 1.
Methods
DATA
Literature derived morphological character matrices were included in this study only when all of the following criteria were
satisfied: (1) stratigraphic ranges were identified for each taxon
(graphically, verbally, that is, with global or regional names of
strata, or numerically); (2) stratigraphic ranges varied among taxa;
(3) at least nine taxa, usable in the current analyses are represented,
to allow a large enough sample to detect trends and to perform rarefaction analyses but not so restrictively large that too few datasets
qualify; (4) taxa are of equivalent taxonomic ranks (exceptions are
noted in Table 1); (5) fossil taxa were represented. Datasets with
many extant taxa are excluded because of the issues of one-sided
range-truncations (see Gilinsky 1988). However, a small number of datasets with partial extant representation were used to increase the sample size and overall taxonomic breadth for this study
(Table 1).
Wagner (2000) assembled a database of morphological character matrices. I searched among these matrices for those matching the criteria listed above. I supplemented Wagner’s collection by initially browsing journals that were represented in
his database (Wagner 2000) and subsequently by systematically searching through those journals that preliminarily yielded
more suitable datasets. These were Lethaia, Historical Biology,
Journal of Paleontology, Paleobiology, and Systematic Biology
(1996–2005). Other journals specializing in systematic studies
yielded surprisingly few datasets that met the listed criteria. More
datasets suitable for the current study were found using references cited in papers that met the listed criteria. Updates of
phylogenetic hypotheses were made for a limited number of
the datasets, but only the most recent paper(s) by the same author(s) discussing the same taxa is/are included here to avoid
duplication. I also included Wagner’s datasets of Paleozoic gastropods (see http://pjw3.fmnh.org/EvolutionMatrices 2000.html)
and used his phylogenetic hypotheses to subdivide some of his
datasets (Table 1). New species character matrices that I coded
from four extinct ostracode genera, namely Curfsina Deroo 1966,
Opimocythere Hazel 1968, Phalcocythere Siddiqui 1971, and
Schizoptocythere Siddiqui and Al-Furaih 1980 were also included
in the analyses (see online Supplementary Materials, Appendices
S1–S3 for character matrices, stratigraphic ranges, character descriptions, and references).
In summary, of the 66 datasets (Table 1) that are used in the
analyses, 38 were used in Wagner (2000). The remaining were
datasets coded for this study (N = 4), Wagner’s own matrices of
Paleozoic gastropods (N = 14), and datasets from other sources
(N = 10). These datasets represent taxa from the Paleozoic (N =
26), the Mesozoic (N = 4), the Cenozoic (N = 20), both the
Paleozoic and the Mesozoic (N = 4), and both the Mesozoic and
the Cenozoic (N = 12). The data span mammals (N = 7), other
vertebrates (N =7), trilobites (N = 4), other arthropods, including
ostracodes, (N = 7), mollusks (N = 27), echinoderms (N = 8),
brachiopods (N = 5), and cnidarians (N = 1) and hence are a broad
representation of fossilizable animals across the Phanerozoic.
DATA TREATMENT
Stratigraphic ranges or geologic durations are explicitly equated to
lineage durations. Henceforth I use these terms interchangeably. It
is assumed that each included study involves closely related taxa
that have similar preservation potentials such that even though
stratigraphic ranges are underestimates of true durations, the rank
order of the ranges reflects the rank order of the true durations
(but see Discussion).
EVOLUTION APRIL 2007
887
888
EVOLUTION APRIL 2007
21
22
19
20
17
18
15
16
13
14
11
12
10
9
7
8
4
5
6
3
2
1
Alvarez et al. 1998
Amati and Westrop
2004
Anderson and
Roopnarine 2003
Angielczyk and
Kurkin 2003
Bloch et al. 2001
Bodenbender and
Fisher 2001
Brochu 1997
Brunet–Lecomte and
Chaline 1990
Cairns 2001
Caron et al. 2004
2004
Damiani 2001
Dashzeveg and
Meng 1998
Dewing 2004
Ebbestad and Budd
2003
Forey 1991
Froelich 2002
Adnet and Capetta
2001
Adrain and Westrop
2001
Adrain and
Edgecomb 1997
Allmon 1996
Allmon 1996
Alroy 1995
Author
some analyses (see text).
Other vertebrates
Mammals
Brachiopods
Trilobites
Cnidarians
Other
arthropods
Other vertebrates
Mammals
Other vertebrates
Mammals
Mammals
Echinoderms
Mammals
Molluscs
Brachiopods
Trilobites
Molluscs
Molluscs
Mammals
Trilobites
Trilobites
Other vertebrates
Group
Sarcopterygii
Equidae
Strophemenata
Burlingiidae
Mastodonsauroidea
Ctenodactyloidea
Dendrophylliidae
Nektaspida
Crocodylia
Terricola
Plesiadapiformes
Blastoidea
Dicynodontia
Corbulidae
Athyridida
Illaenidae
Turritellidae
Turritellidae
Hipparionini
Encrinurine
Ptychaspididae
Squaliformes
AF
F
AF
F
AF
AF
F
AF
AF
G
AF
AF
AF
F
AF
G
F
F
SB
SB
F
AF
Domain
G
S
S
S
G
G
G/SG
G/S
S
S
S
G
G
G/S
F/SB
S
G/SG
S
S
S
S
G
Unit
31
14
9
16
21
17
30
9
61
16
14
68
20
12
36
19
51
36
17
31
12
23
N
56
47
15
19
38
26
10
12
164
3
32
94
53
70
37
17
14
30
56
40
16
29
Nchar
Scythian–Recent
Eocene
Ashgill–Llandovery
Mid-upper Cambrian
Cretaceous–Recent
Cambrian–
Ordovician
Permian–Triassic
Eocene–Miocene
Cretaceous–Recent
.5–0 MYA
Paleocene–Eocene
Llandeilo–Kazanian
Kazanian–Anisian
Cretaceous–Recent
Ordovician–Jurassic
Mid–late Ordovician
Late Cretaceous–Recent
Paleocene–Eocene
Miocene–Pliocene
Telychian–Ludfordian
Sunwaptan–Ibexian
Late Jurassic–Pleistocene
Geologic range
M
T
M
M
T
T
M
M
M/FW
T
T
M
T
M
M
M
M
M
T
M
M
M
Realm
MY
L
L
S
MY
MY
MY
MY
MY
MY
S
S
S
MY
MY
S
MY
MY
MY
S/L
S
MY
DUR
N
N
N
Y
Y
Y
N
Y
N
N
Y
Y
Y
N
Y
N
N
N
Y
Y
Y
Y
DUR-I
continued
Inferred
durations only
Teeth only
[OL 4]
Inferred durations
only
Teeth only
Notes
some notes mentioned in the text. In particular, parentheses indicate those datasets overlapping with other datasets (OL N = overlapping with dataset number N from column 1) that were removed for
whether the durations (DUR) are measured in millions of years (MY), stages (S), or manually measured on range charts (L), and finally if inferred durations (DUR-I) were available. The last column records
(N), the number of characters used in the analyses (Nchar), the geologic range represented in the studies, the biological realm in which the clades are found (M = marine, FW = freshwater, T = terrestrial),
of the studies (where AF = above family, F = family, SB = subfamily, SG = genus, G = genus), the taxonomic unit whose characters were coded (as before, with S = species), the number of taxa involved
Table 1. The references used in the analyses, the groups they represent (a general grouping followed by the latin name of the domain or the higher level grouping encompassing the domain), the domain
LEE HSIANG LIOW
39
38
34
35
36
37
30
31
32
33
29
27
28
25
26
24
23
Continued.
Gahn and Kammer
2002
Grande and Bemis
1998
Hopkins 2004
Jeffery and Emlet
2003
Jeffery 1998
Karasawa and
Kato 2003
Leighton and
Maples 2002
Michaux 1989
Monks 1999
Monks 2002
Monks and Owen
2000
Nutzel et al. 2000
O’Keefe 2004
Popov et al. 1999
Roopnarine
2001–2001
Roopnarine
2001–2002
Roopnarine
2001–2003
Author
Table 1.
Molluscs
Molluscs
Molluscs
Other vertebrates
Brachiopods
Molluscs
Molluscs
Molluscs
Molluscs
Brachiopods
Brachiopods
Echinoderms
Other arthropods
Mammals
Echinoderms
Other vertebrates
Echinoderms
Group
Chione
Puberella
Subulitoidea
Sauropterygia
Atrypida
Chione
Ancillinae
Ancylocertina
Hamitidae
Orbirhynchia
Productida
Cyclaster
Goneplacidae
Rodentia
Temnopleurids
Amiidae
Botryocrinidae
G
G
AF
AF
AF
G
SB
AF
F
G
AF
G
F
G
AF
F
G
Domain
S
S
G
G/S
S
S
S
S
S
S
G
S
G
S
S
S
S
Unit
13
17
11
12
25
16
20
25
23
16
14
10
15
9
16
22
10
N
13
20
16
88
27
20
36
26
30
22
24
22
45
30
38
46
14
Nchar
Oligocene–Recent
Oligocene–Recent
Devonian–Triassic
Jurassic
Ordovician
Oligocene–Recent
Eocene–Recent
Lower Albian–upper Albian
Lower Albian–upper Turonian
Albian–Campanian
Givetian–Pennsylvanian
Late Cretaceous–Paleogene
Paleogene–Recent
38–15 MYA
Eocene–Pliocene
Cretaceous–Recent
Mississippian
Geologic range
M
M
M
M
M
M
M
M
M
M
M
M
M
T
M
M/FW
M
Realm
MY
MY
MY
MY
L
MY
MY
MY
MY
MY
S
S
MY
MY
MY
MY
L
DUR
N
N
Y
Y
N
N
N
Y
Y
Y
N
N
N
Y
N
N
N
DUR-I
continued
[OL 37]
[OL 32]
Teeth only
Notes
MORPHOLOGY AND LINEAGE DURATIONS
EVOLUTION APRIL 2007
889
Continued.
890
EVOLUTION APRIL 2007
Wagner (2000)
Wagner (2000)
Wagner (2000)
Wagner (2000)
Wagner (2000)
Wagner (2000)
Wagner (2000)
Yates and Warren
(2002)
Liow
Liow
Liow
Liow
55
56
57
58
59
60
61
62
63
64
65
66
40 Roopnarine
2001–2004
41 Schneider 1995
42 Smith 1988
43 Smith and Arbizu
1987
44 Smith et al. 1995
45 Smith and Wright
1993
46 Tinn and Meidla
2004
47 Vermeij and
Carlson 2000
48 Wagner 1999
49 Wagner 1997
50 Wagner 1997
51 Wagner 1997
52 Wagner 1997
53 Wagner 1997
54 Wagner (2000)
Author
Table 1.
Cardiidae
Molluscs
Echinoderms
Echinoderms
Rapaninae
Lophospiroida
Rostrochoncha
Ribeiriidae
Technophoridae
Bransoniidae
Hippocardiidae
Paleozoic
gastropods
Euomphaloid
Pleurotomarioid
Trochoid
Murchisonioid
Microdomatoid
Trochonematoid
Macluritoid
Temnospondyli
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Ostracodes
Ostracodes
Ostracodes
Ostracodes
Curfsina
Opimocythere
Phalcocythere
Schizoptocythere
Beyrichiocopa
Ostracodes
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Molluscs
Other
Vertebrates
Ophiuroidea
Echinoderms
Echinoderms
Agelacrinitinae
Puberella
Molluscs
Group
G
G
G
G
AF
AF
AF
AF
AF
AF
AF
AF
AF
AF
F
F
F
F
AF
SB
AF
AF
AF
F
AF
SB
G
Domain
S
S
S
S
S
S
S
S
S
S
S
G
S
S
S
S
S
S
S
G/S
S
SB
G
G/SG
G
G
S
Unit
29
17
30
16
67
202
13
66
12
15
15
34
82
154
27
17
22
39
481
7
16
8
9
146
167
85
107
61
57
67
60
91
126
46
62
50
68
217
36 34
35 39
28 41
14 29
32 16
29 32
13 12
Mid Albian–Thanetian
Upper Albian-mid Miocene
Upper Maestrichtian—Oligocene
Lower Santonian–mid Miocene
Early Tremadoc–Eifelian
Early Tremadoc–Eifelian
Llanvirn–early Ludlow
Early Arenig–Eifelian
Early Caradoc–late Ludlow
Early Caradoc–Late Ludlow
Mid-Cambrian–Ashgill
Carboniferous–Jurassic
Cassinian–Pridoli
Early Cambrian–Capitanian
Early Cambrian–upper Caradoc
Early-mid Cambrian–Ashgill
Upper Arenig–upper Caradoc
Llanvirn–Serpukhovian
Early Cambrian–Givetian
Eocene–Recent
Early–Middle Ordovician
(Permian) Triassic– Recent
Jurassic–Recent
Triassic–Recent
Ordovician–Carboniferous
Ordovician–Carboniferous
Oligocene–Recent
Nchar Geologic range
15 19
N
M
M
M
M
M
M
M
M
M
M
M
T
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
MY
MY
MY
MY
MY
MY
MY
MY
S
MY
MY
MY
S
S
S
S
S
S
MY/S
MY
S
MY
MY
MY
MY
S
MY
N
N
N
N
N
N
N
N
N
N
N
Y
N
N
N
N
N
N
N
N
N
Y
Y
Y
N
N
N
Families deleted
[OL 55–61]
[OL 50–53]
[OL 38]
Realm DUR DUR-I Notes
LEE HSIANG LIOW
MORPHOLOGY AND LINEAGE DURATIONS
The data treatment here is similar to two previous analyses of
lineage duration versus morphological distributions (Liow 2004,
2006), but with two crucial improvements. First, long-duration
lineages are dynamically defined, not static subsets of the lineages in question. Second, long-duration lineages are compared
with short-duration lineages in groups (group analyses) as well as
individually (single analyses) (see below and Fig. 2).
The relative morphological distance of each taxon in a given
dataset is calculated as the sum of the distance of each of its
character state from each corresponding average character state
of the dataset. Both unweighted distances and weighted distances
where each character contributes equally to the total distance of
a taxon from the average of the given dataset were calculated.
An average character state is calculated as either the modal or
mean character state. The latter is reasonable for binary and ordered multistate characters, but less so for unordered multistate
characters. Hence unordered multistate characters are converted
into binary characters by coding the modal character state as “0”
and all other character states as “1.” Numerical or ordered multistate characters having character states with a minimum of “6”
are log transformed (new value = ln (old value+2)) so that they
will not dominate the calculation of morphological distances in
unweighted treatments.
Some of the datasets included here were originally assembled for cladistic analyses. The exceptions are the four ostracode
species datasets coded for this study and Wagner’s Paleozoic
gastropod datasets used for analyses of morphospace occupation as well as for phylogenetic analyses. There may be a concern that cladistic datasets represent a biased sample of morphological characters, relative to phenetic morphological character
datasets (e.g., Foote 1999; Liow 2006). However, the cladistic
nature of the datasets does not affect my analyses for the following reasons. Cladistic datasets consist of an array of presumably
evolutionarily significant pleisiomorphic and derived characters
obtained from a sampling of the morphology of the organisms
they describe. Cladistic data are hence valuable in this attempt
to capture relative morphological distributions. In the current
analyses, outgroup taxa used to polarize cladistic analyses are
removed as they may only be relatively distantly related to the
inclusive group and hence can skew the calculation of the grand
empirical morphological average. In doing so, some characters
in the ingroup taxa may reflect only one state such that they
do not contribute to the calculated distance matrices. Likewise,
when larger groups are parsed into smaller groups for further
analyses, some characters become uniformly represented. These
uniform characters are removed from the calculations because
they will add to the counts of characters used (Table 1) without contributing to the distance measures described here. The
purposeful exclusion of autapomorphies from cladistic datasets
may bias results if some lineages have many of them, that is, if
a lineage is actually very deviant morphologically from an average because of their large number of unique characters. Two
factors counteract this potential bias: (1) the analyses of each
dataset involve, in part, relative differences computed in Principal Component space (see later section) such that autapomorphies do not contribute to principal axes and hence also do not
contribute to the results; (2) autapomorphies will only systematically bias results calculated using Euclidean type distance metrics,
if they are distributed nonrandomly with respect to durations.
However, there is no prior expectation that this biased distribution exists.
Three types of stratigraphic ranges of taxa are reported in the
literature (Table 1). First and most commonly, numerical values or
stage names were given by the authors of the papers (MY, Table 1).
Ranges from the prior were used directly as durations; the latter
were converted to numerical values of the midpoints of the geologic stages using Gradstein et al. (2004). Second, where time
intervals not conforming to internationally recognized names were
reported, numbers sequentially assigned to reflect their chronological order were used to calculate relative durations (S, Table 1).
Finally, if no stage names or durations are given, I measured the
illustrated lengths of stratigraphic ranges if they were drawn to
scale (L, Table 1). Where values were calculated, durations have
a value of zero if a taxon is found only in one named stage (i.e.,
a singleton taxon). Because rank order statistics are used in this
study, allowing singletons to have equal rank order durations is
preferable to allowing singletons to have varying rank order durations stemming from differences in the lengths of named stages.
These differences in stage length can artificially introduce certainty in duration variation of singletons. In cases where more than
one of these types of stratigraphic ranges was available, the data
are analyzed with each of them to check for possible differences
in results.
Inferred geologic range extensions based on phylogenetic
inference are reported in some studies (e.g., Bloch et al. 2001;
Bodenbender and Fisher 2001). Where such inferences were available (Table 1), I also analyzed these data with the inferred durations (see Lane et al. 2005 for a discussion of types and properties
of various inferred durations).
In group analyses, taxa with long durations are increasingly
inclusive sequential groups (Fig. 2B). First, a long-duration group
is simply the taxon with the longest duration (taxon A in Fig. 2B),
then the two taxa with the longest durations (taxa A and B in
Fig. 2B), then the three taxa with the longest durations (taxa A,
B, and C in Fig. 2B) and so on, until half the taxa have been
included in the long-duration group. Then I compare the mean
relative morphological distance of each long-duration group with
that of an equivalent number of randomly selected short-duration
taxa, with replacement. This rarefaction is done because sample
sizes are different for taxa with longer and shorter durations (see
EVOLUTION APRIL 2007
891
LEE HSIANG LIOW
A
Morphological distance from the grand empirical average
B.1
Morphological distance from
the grand empirical average
also Liow 2004, 2006). The rarefaction was repeated 500 times to
determine the frequency with which a given long-duration group
has a mean relative morphological distance that is less than the
mean calculated for each randomly selected short-duration group.
This bootstrap-generated frequency corresponds to the probability
of obtaining a long-duration group with a relative morphological
distance that is less than or equal to a random draw of an equivalent
G
number of morphologies of short-duration taxa. The procedure
was then reversed such that the probability of obtaining a shortduration group that is more distant from the grand average than
observed is tabulated. Combining both sets of probabilities and
plotting them against their respective group durations results in a
morpho-duration distribution plot (Fig. 1), for which a rank order
correlation (Kendall’s tau) and corresponding P-value (p(g)) was
C
E
F
A
B
Lineage duration
Group Analysis
C.1
Single Analysis
C
C
A
A
B
B
B.2
C.2
C
C
A
A
B
B
B.3
C.3
C
C
A
A
B
B
Lineage duration
Figure 2. Panel A is a hypothetical plot showing the morphological distance of each taxon (black circles) from the grand empirical
average plotted versus its lineage duration. The plots in panels B and C are replicas of panel A. Panel B illustrates a group analysis where
sequentially larger groups of taxa (A, A + B, A + B + C, etc.) with longer durations are compared with rarified samples of the remaining
taxa of shorter durations to the left of the circled taxa (Fig. 2B.1 through B.3). Panel C illustrates a single analysis where individual taxa (A,
then B, then C, etc.) are compared with a randomly picked taxon from the remaining pool of taxa (excluding taxa with longer durations
than the one being compared) with shorter durations (Fig. 2C.1 through C.3). Taxa G, E, and F in panel A have the same durations and
their median probabilities of being more or less distant from the grand average morphology are used in subsequent analyses for both
group and single analyses.
892
EVOLUTION APRIL 2007
MORPHOLOGY AND LINEAGE DURATIONS
calculated to determine which of the scenarios illustrated in Figure
1 best describes the dataset in question (tau and p(g) are reported
in online Supplementary Materials, Appendix S4).
Various taxa with long durations may be individually morphologically less distant from the grand average than expected,
when compared with their short-duration relatives. Single analyses were done by sequentially defined long-duration taxa
(Fig. 2C). Each long-duration taxon was compared with a randomly selected short-duration taxon, that is, excluding taxa that
have longer durations than the long-duration taxon in question.
This was repeated 500 times. This bootstrap-generated frequency
corresponds to the probability of obtaining a long-duration taxon
with a relative morphological distance that is less than or equal
to the random draws of morphologies of short-duration taxa. The
procedure was then reversed such that the probability of obtaining
a short-duration taxon that is more distant from the grand average
than observed is tabulated. As described for group analyses, a rank
order correlation (Kendall’s tau) and corresponding P-value (p(s))
were then calculated for the combined set of probabilities versus
their respective durations. The sign and probability of the correlation were used to determine which of the scenarios illustrated
in Figure 1 best describes the dataset in question (tau and p(s) are
reported in the online Supplementary Materials, Appendix S4).
In both group and single analyses, I repeat the procedures described above but replace relative morphological distances with
principal component scores obtained from Principal Component
Analysis of distance matrices obtained using the character matrices (Principal Coordinate Analyses [PCO] Gower 1966). This
data reduction technique removes redundancy in the original morphological data. The number of scores used is adjusted to explain
about 80% of the variance and varies from five to 20, depending
on the size of the data matrix. This bootstrap-generated frequency
corresponds to the probability of obtaining a long-duration group
with a sum of principal component scores that is less than or
equal to the same of a random draw of an equivalent number of
morphologies of short-duration taxa. These probability values for
group and single analyses are plotted against taxon durations and
rank order correlations (Kendall’s tau) and their associated probabilities for group and single analyses (p(g, pco) and p(s, pco),
respectively) are calculated (see online Supplementary Materials,
Appendix S4).
Taxa with the same calculated durations could have different
bootstrapped probabilities of being morphologically distant from
the grand average (e.g., taxa E, F, and G in Fig. 2A). I recorded the
median bootstrapped probabilities of taxa with the same calculated
duration. These are then used in performing rank order correlations
that indicate whether long-duration taxa are less distant from the
grand average morphology than by chance alone in both group and
single analyses. The abbreviations p(g, m), p(g, m, pco), and p(s,
m), p(s, m, pco) are used to indicate the probabilities from the rank
order correlations of median bootstrapped probabilities of group
analyses, group analyses using PCO values, single analyses, and
single analyses using PCO values versus durations, respectively
(online Supplementary Materials, Appendix S4).
There are multiple ways of quantifying stratigraphic ranges
and relative morphological distances, that is, using original and inferred durations, stratigraphic ranges measured in different ways,
using the mode or mean character states as the average, or distances or principal coordinate scores. At the minimum, there are
8 combinatory ways of calculating morpho-duration distributions
(the Brunet-Lecomte and Chaline 1990 dataset, see Table 1 and
online Supplementary Materials, Appendix S4) and at maximum
there are 48 ways (the Adrain and Edgecomb 1997 dataset, see
Table 1 and online Supplementary Materials, Appendix S4). Correspondingly, significance tests of trends or mopho-duration distributions (Fig. 1) may differ among various data treatments within
a dataset. I present all the results obtained (online Supplementary
Materials, Appendix S4) using various data treatment variants but
summarize whether the taxa represented in a given dataset show a
positive, negative, or null morpho-duration distribution using the
following criteria. If there is only one significant result in the possible data variants, then the dataset is assumed to demonstrate a null
morpho-duration distribution (Fig. 1C). If opposing significant results are in a ratio of one to one, then the relationship is also taken
to be nonsignificant. If opposing significant results are in a ratio of
more than one to one, then the more commonly represented sign is
accepted. For instance, if there are three significantly negative values and only one significantly positive value among the data variants of a given dataset, then this dataset is taken to show a negative
morpho-duration distribution (Fig.1B). In addition, I consider the
possibility that any conflict of the signs of significant correlation
indicates a nonsignificant situation and refer to this as a conservative solution (see Results). Because significant cases are sometimes already removed in this method of summary, Bonferroni
corrections that further overcorrect for significance are not used.
The datasets used for these analyses differ in their focal taxonomic level. To test whether the difference in taxonomic scale affects conclusions drawn regarding morpho-duration distributions,
I first enumerated Linnean taxonomic ranks of the taxa whose
morphologies are coded (coded taxon value) as well as that of the
domain of the study (domain taxon value), such that 0 = species,
1 = subgenus, 2 = genus, 3 = subfamily, 4 = family, and 5 =
above family. I then calculated the taxonomic inclusiveness of
each study as domain taxon value minus the coded taxon value.
As an illustration, Jeffery and Emlet (2003) studied temnopleurid
echinoids (domain taxon value = 5) and coded the morphology of
temnopleurid species (coded taxa value = 0), hence the taxonomic
inclusiveness is 5. Where there is a mixture of ranks of taxa whose
morphologies are coded, I use the mean value of the coded taxon
values (e.g., if both genera and subgenera were coded, then the
EVOLUTION APRIL 2007
893
LEE HSIANG LIOW
This table groups references listed in Table 1 according to whether they demonstrate a positive, NEGative, or null morphoduration distribution (see Fig. 1) in group analyses (GROUP). Results of single analyses (SINGLE) follow. In addition, (raw) mean and
median durations (M.y.), standard deviation of durations, beta, alpha, and skew are listed. Durations and standard deviations thereof
are calculated only for those datasets where stratigraphic ranges are reported in or convertible to millions of years. An asterisk implies
that conservatively, the diagnosis would have reflected a null morpho-duration distribution.
Table 2.
Study
GROUP
Adrain and Edgecomb 1997
Allmon 1996 (Table 9)
Alvarez et al. 1998
Anderson and Roopnarine 2003
Angielczky and Kurkin 2003
Bloch et al. 2001
Brochu 1997
Brunet-Lecomte and Chaline 1990
Cairns 2001
Damiani et al. 2001
Gahn and Kammer 2002
Jeffery 1998
Leighton and Maples 2002
Monks 2002
Monks and Owen 1999
Opimocythere (this study)
Phalcocythere (this study)
Popov et al. 1999
Roopnarine 2001 (set 1)
Wagner 1997
Wagner 1999
Wagner 2000 (Euomphaloid)
Wagner 2000 (Macluritoid)
Wagner 2000 (Microdomatoid)
Wagner 2000 (Mursonoid)
Wagner 1997 (Ribeiriidae)
Wagner 1997 (Technophoridae)
Wagner 1997 (Bransoniidae)
Wagner 1997 (Hippocardiidae)
Wagner 2000 (Trochoids)
Wagner 2000 (Trochonematoid)
Schneider 1995
Adrian and Westrop 2001
Alroy 1995
Amati and Westrop 2004
Caron et al. 2004
Curfsina (this study)
Dashzeveg and Meng 1998
Dewing 2004
Ebbestad and Budd 2003
Froelich 2002
Grande and Bemis 1998
Hopkins 2004
Jeffery and Emlet 2003
Karasawa and Kato 2003
Monks 1999
Nutzel et al. 2000 set-2
NEG
NEG
NEG∗
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG
NEG∗
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
SINGLE
Mean-Dur
Med-Dur
Sd-dur
Beta
Alpha
Skew
NEG
NS
POS∗
NEG
NS
NS
NEG
NEG
NEG
NS
NS
NS
NS
NEG
NEG
NS
NS
NS
NS
NEG
NEG
NS
NEG
NEG
NS
NEG
NEG
NS
NEG
NS
NEG
NS
NS
NS
NS
POS
NS
NS
NS
NS
NEG
POS
POS
NS
NS
NS
NS
NA
3.7
43.5
27.8
NA
NA
2.9
.2
16.9
2.9
NA
NA
NA
1.4
2.3
5.3
3.1
NA
1.0
NA
NA
5.2
2.1
1.9
4.1
NA
NA
NA
NA
6.2
2.9
88.1
NA
1.6
1.9
22.7
3.6
2.0
NA
NA
.7
NA
2.2
6.3
13.6
.5
59.5
NA
3.2
33.5
14.8
NA
NA
1.8
.1
5.9
.0
NA
NA
NA
.0
.0
1.5
.6
NA
.0
NA
NA
2.6
.0
.0
.0
NA
NA
NA
NA
6.1
.0
95.0
NA
1.0
.9
14.5
2.0
.0
NA
NA
.7
NA
1.5
4.0
.0
.0
62.0
NA
2.3
42.9
28.2
NA
NA
3.8
.2
23.7
4.9
NA
NA
NA
2.3
3.2
8.6
5.4
NA
1.5
NA
NA
6.4
3.0
2.8
6.9
NA
NA
NA
NA
10.6
4.1
54.0
NA
.9
2.1
34.6
5.8
3.2
NA
NA
.4
NA
1.8
4.4
17.0
1.4
57.4
.7
.6
1.0
1.0
1.7
1.1
1.3
.8
1.4
1.7
1.1
.7
1.2
1.6
1.4
1.6
1.7
.5
1.4
2.3
1.1
1.2
1.5
1.5
1.7
1.4
1.0
3.0
2.8
1.7
1.4
.6
1.8
.6
1.1
1.5
1.6
1.6
.6
.8
.6
2.0
.8
.7
1.3
2.8
1.0
1.7
6.0
44.1
27.5
.4
.5
2.3
.3
12.0
1.7
1.8
5.2
.6
.9
1.7
3.3
1.8
1.8
.7
.1
39.4
4.3
1.4
1.3
2.4
.4
1.0
.1
.1
3.6
2.1
143.8
.3
2.7
1.7
14.9
2.2
1.2
4.2
1.9
1.1
1.2
2.7
9.0
10.8
.2
61.6
1.5
.8
.3
.4
3.0
2.8
1.3
3.7
.6
1.5
1.5
.9
2.7
2.1
1.6
1.1
1.5
1.5
2.3
5.7
.3
1.0
1.7
1.8
1.3
3.2
2.0
7.3
7.4
1.1
1.4
.2
3.8
1.2
1.5
.5
1.3
1.8
1.0
1.4
1.9
1.8
1.2
.7
.6
4.7
.3
continued
894
EVOLUTION APRIL 2007
MORPHOLOGY AND LINEAGE DURATIONS
Table 2.
Continued.
Study
GROUP
Schizoptocythere (this study)
Smith 1988
Smith et al. 1995
Vermeij and Carlson 2000
Wagner 2000 (Paleozoic coiled gastropods)
Wagner 2000 (Pleurotomarioid)
Adnet and Capetta 2001
Allmon 1996 (Table 1)
Bodenbender and Ficher 2001
Forey 1991
Michaux 1989
O’Keefe 2004
Roopnarine 2001 (set 2)
Roopnarine 2001 (set 3)
Roopnarine 2001 (set 4)
Smith and Arbizu 1987
Smith and Wright 1993
Tinn and Meidla 2004
Yates and Warren 2002
NS
NS
NS
NS
NS
NS
POS
POS
POS
POS
POS
POS
POS
POS
POS
POS
POS
POS
POS
SINGLE
Mean-Dur
Med-Dur
Sd-dur
Beta
Alpha
Skew
NS
NS
NEG
NS
POS
NS
NS
NS
POS
NS
NS
POS
NS
NS
NS
NS
NS
NS
POS
2.7
3.2
59.6
11.4
3.9
4.7
64.3
7.7
NA
7.4
5.2
1.3
1.0
1.2
.9
NA
29.6
NA
8.7
.0
.0
33.5
10.0
.0
.0
62.3
.0
NA
.0
4.2
.0
.0
.0
.0
NA
18.0
NA
.0
6.0
10.1
70.2
12.3
6.7
7.5
45.2
16.0
NA
17.6
5.2
2.6
1.5
1.6
1.3
NA
33.7
NA
13.6
2.2
3.2
1.2
1.1
1.7
1.6
.7
2.1
1.5
2.4
1.0
2.0
1.4
1.3
1.4
1.4
1.1
.9
1.6
1.3
1.0
50.8
10.3
2.2
3.0
91.3
3.7
.6
3.1
5.2
.7
.7
1.0
.6
2.5
26.0
.8
5.1
1.8
2.0
.3
.6
1.3
1.2
.2
1.0
2.5
1.1
.9
2.5
2.3
2.0
2.5
1.3
.4
2.2
.9
coded taxon value = 1.5). I also tabulated the number of morphological characters used in my analyses, the number of taxa, and
whether the group in question is from the aquatic or terrestrial
realm. In addition, I calculated mean and median durations for
taxa that have stratigraphic ranges reported in millions of years,
the standard deviation of the durations, alpha (square of the mean
divided by standard deviation of durations) which is a shape parameter, beta (standard deviation divided by square root of alpha)
which is a scale parameter and skew (two divided by square root
of alpha) (Wackerly et al. 2002). The latter three are descriptors
of gamma distributions, which I assume are approximated by the
duration distributions of the datasets because durations are always
nonnegative and right-skewed.
For completeness, I also reanalyzed the data used in Liow
(2004, 2006) to compare previous results using this newly developed continuous method of comparing morphological distance
versus durations.
Results
First, a few statements are made to make way for economy
and clarity. Weighted and unweighted relative morphological distances do not provide different results in the morpho-duration distributions. Hence, I discuss only the results using weighted relative
morphological distances. Similarly, using either grand empirical
average morphologies of inclusive groups calculated as sums of
mean or modal character states did not in general change resulting patterns of morpho-duration distributions. Likewise, the use of
different approaches of quantifying durations did not consistently
result in qualitatively different probabilities for the same datasets.
Finally, the frequentist approach of determining which particular
pattern of morpho-duration distribution best represents the situation in a given dataset (see Methods) removes inconsistencies
(see online Supplementary Materials, Appendix S4 for detailed
results).
A majority of the 66 datasets show a significant negative
morpho-duration distribution (Fig. 1B), but positive distributions
(Fig. 1A) and null distributions are also represented (Fig. 1C,
Tables 2 and 3). Because some datasets were subsets of others
or have overlapping taxa (Table 1), I retabulated the number of
cases of each of the above excluding the overlaps and found that
the rank order of the number of studies of each distribution type
was not altered (Table 3). Thus, when taxa are widely sampled, all
three of the described scenarios of morpho-duration distributions
are observed. In the majority of the cases, relative morphological
distances are negatively distributed with respect to durations (a
negative morpho-duration distribution). Even using a conservative
approach, where a dataset is considered to have a nonsignificant
relationship when the signs of the correlations disagree among
treatments, datasets that show a negative morpho-duration distribution still predominate (Table 2). Notice that grand means of rank
order correlations (tau) do not change in sign when nonsignificant
results are used in their calculation (Table 3).
In group analyses, datasets showing negative, positive, and
null morpho-duration distributions neither have significantly
different numbers of taxa represented nor different numbers of
coded morphological characters (t-test, P > 0.05).
EVOLUTION APRIL 2007
895
LEE HSIANG LIOW
Table 3. This is a summary of the results from Table 2 and Online Appendix 4 for GROUP and SINGLE analyses. The first row in each block
(positive, negative, and nonsignificant morpho-duration distributions) shows the percentages of datasets that display the respective
morpho-duration distributions. Percentages in parentheses exclude potentially overlappping datasets (Table 1). Mean Tau is the grand
mean rank order correlation followed by its standard deviation; mean P is the grand mean P value followed by its standard deviation,
calculated from the significant cells of each dataset (online Supplementary Materials, Appendix S4). The numbers following in brackets
are the same metrics averaged from all the cells in each dataset grouping.
Morpho-duration distribution
Positive
Negative
Nonsignificant
Percentage
Mean Tau
Mean P
Percentage
Mean Tau
Mean P
Percentage
Mean Tau
Mean P
Datasets demonstrating a positive morpho-duration distribution have marginally significantly greater coded taxon values (0.8)
than either those demonstrating a negative one (0.3, t-test, P =
0.09) or a null one (0.3, t-test, P = 0.05). There is no significant
difference in either their domain values or taxonomic inclusiveness (t-test, all P > 0.05).
Datasets representing organisms from aquatic (marine and
freshwater) or terrestrial environments are not differentially represented in the three patterns of morpho-duration distributions
(chi-square test, all P > 0.05).
In terms of taxonomic representation, there is a significant
difference in terms of distribution of mammal representation (chisquare test, P = 0.03). Datasets demonstrating a negative morphoduration distribution are represented by three mammal studies out
of a total of 32, no null cases out of a total of 21 and groups having
a positive distribution are four out of a total of 13. Other common
groups represented in the comparisons, mollusks, echinoderms,
and all vertebrates considered together, show no significant differences in frequencies among the three patterns of morpho-duration
distribution.
Whether the datasets represent only extinct or both extinct
and extant lineages, is also not a factor in their distribution among
the three patterns of morpho-duration distributions (chi-square
test, all P > 0.05).
Datasets showing negative, positive, and null morphoduration distributions do not have significantly different mean
or median durations, or descriptors of the duration distribution,
including the shape parameter alpha, the scale parameter beta and
skew (t-tests, all cases P 0.05).
The former results are for groups of increasingly inclusive
long-duration and short-duration taxa. In contrast, for comparisons of individual taxa with the remaining longer duration or
shorter duration taxa in single analyses, most datasets demon-
896
EVOLUTION APRIL 2007
GROUP
SINGLE
20 [18]
.47 ± .05 [.34 ± .10]
.01 ± .01 [.26 ± .13]
48 [50]
−.49 ± .09 [−.31 ± .12 ]
.01 ± .01 [.27 ± .13]
32 [32]
−.08 ± .29 [−.04 ± .13]
NA [.36 ± .08 ]
12 [10]
.39 ± .10 [.24 ± .22]
.02 ± .01 [.30 ± .28]
26 [27]
−.46 ± .06 [−.30 ± .29 ]
.01 ± .01 [.26 ± .26]
62 [63]
−.17 ± .02 [−.01 ± .21]
NA [.50 ± .28 ]
strated the null morpho-duration distribution. In fewer of the
cases, longer duration taxa are individually morphologically
closer to the grand average than expected, demonstrating a negative morpho-duration distribution, and in rarely do they have
a positive morpho-duration distribution (Table 3). Even after
disregarding cases that may be nonindependent, because they
stem from studies using overlapping taxa, the percentages of
occurrence of each morpho-duration distribution remain similar
(Table 3).
Datasets consisting only of extinct taxa have lineage durations that are significantly positively correlated with age of the
lineages in 16 out of 47 cases (Kendall’s rank test, P < 0.05). The
31 nonsignificant cases show positive correlation coefficients in
all but six cases, each with very small negative coefficients (data
not shown). Datasets including extant taxa have lineage durations
that are significantly positively correlated with age of the lineages
in 13 out of 19 cases (Kendall’s rank test, P < 0.05), with the
remaining six being nonsignificant (data not shown).
Analyses using crinoid genus data (Liow 2004) corroborate
the prior results. The crinoid orders where discrete groups of
longer duration genera were found to be morphologically less
distant from an average than expected in the previous study
demonstrated significant negative morpho-duration distributions
in this study. In addition, individual compared longer duration taxa
(single analyses) demonstrate mainly negative and null morphoduration distributions (Online Appendix 4).
Group analyses using trachyleberidid ostracode data (Liow
2006) proved different in some cases in comparison to the conclusions drawn previously (Online Appendix 4). In particular,
contemporaneous genera and genera in cohorts originating in the
same geologic stage or time interval that were previously thought
to show a positive relationship between morphological distance
and duration show a negative morpho-duration distribution in this
MORPHOLOGY AND LINEAGE DURATIONS
new analysis using a dynamic definition of long-duration forms
(Online Appendix 4).
Discussion
Much has been written about stasis at the species level from the
view points of paleontology, population biology, genetics, and development (Van Valen 1982; Wake et al. 1983; Rutherford 2000;
Merilä et al. 2001; Schwenk and Wagner 2001; Belade et al. 2002;
Eldredge et al. 2005; Grether 2005). However, proposed mechanisms maintaining within-population or within-species phenotypic stability do not inform us of the morphological patterns
of lineage durations among species and higher-than species lineages. Quantitative comparisons of persistent morphologies are
not straightforward because lineages that have comparatively long
durations are relatively rare (small sample sizes).
The present study is a continued attempt to quantify the relationship between morphology and duration within a rigorous
framework. Taxonomic and methodological limitations of two
previous attempts (Liow 2004, 2006) have been alleviated. A
newly developed sequential rarefaction approach allows durations
to be treated as a continuum in analyses relating them to morphologies. Moreover, lineages are also treated as either individually
having long durations, or persistent as groups.
Lineages with long durations could conceivably be morphologically more distant from the grand average of their inclusive
group than under random expectations. This scenario may indicate
that being different confers a competitive edge (Fig. 1A, a positive
morpho-duration distribution). Conversely, lineages with long durations could be morphologically closer to the grand average than
expected (Fig. 1B, a negative morpho-duration distribution). This
second scenario may indicate that being average confers flexibility and generality. What do we observe from studying a large
suite of independently collected data representing diverse groups?
When lineages with longer durations are considered collectively
(group analyses), both of the described scenarios as well as the
null situation were observed (Fig. 1). However, a greater number
of cases display a negative morpho-duration distribution (Table 3),
including the genera of crinoid orders reanalyzed using the data
from Foote (1999) in Liow (2004), as well as the genera of a large
family of ostracodes (using Liow 2006, see online Supplementary
Materials, Appendix S4). In contrast, when these long-duration
lineages are individually considered (single analyses), null
morpho-duration distributions are predominant (Fig. 1C, Table 3).
EVOLUTIONARY IMPLICATIONS
Datasets showing group positive morpho-duration distributions
are more likely to reflect the morphology of genera or families;
those showing null or negative group distributions are more likely
to reflect the morphology of species or subgenera. A positive
morpho-duration distribution suggests that sufficiently distinct
morphologies identified as higher level (para)clades that are morphologically deviant from equivalent groups, persist for longer
periods of time. Perhaps distinct or deviant morphologies could
invade new niches, that is, niches previously unused by related
groups, thus encountering minimum competition, allowing for
the production of descendents which perpetuate the lineage possessing those morphologies. However, positive morpho-duration
distributions are less common than negative ones hence much of
the remaining discussion will ruminate upon the latter distribution.
When less inclusive lineages in the biological hierarchy (e.g.,
species or subgenera) are studied in group analyses, the morphoduration distributions observed are frequently negative. This suggests that being similar to morphological forms that have already
proven effective for closely related lineages may help promote
survivorship or persistence. This may also reflect an above-theindividual-level analogue of genetic compensation, where selection is expected to favor genetic changes that restore the ancestral
phenotype, because any change in development caused by perturbations are likely to reduce fitness (Grether 2005).
The observation that group analyses of taxa shows significant negative or positive morpho-duration distributions more frequently than individually considered taxa may reflect one of two
things. The first less interesting possibility is that there is more
statistical power in bigger numbers (i.e., comparing groups rather
than individuals). The second possibility is that taxa with longer
durations are begetting descendents that are not only morphologically similar to themselves (pulling the grand average morphological value closer to their own morphology) but that also
tend to have long durations. This idea may hold for two reasons.
First, higher-than-organismal-level properties are heritable (e.g.,
Jablonski and Hunt 2006). Second, there is at least a 10% probability of finding ancestors in the fossil record (Foote 1996). Even
if direct ancestors are not present in the datasets analyzed, very
closely related lineages may also produce the same patterns. A
variation to this second possibility is that long-duration lineages
do not necessarily give rise preferentially to long-duration lineages, but rather, are prolific. In their proliferation, they give rise
to both longer and shorter duration descendent lineages, simply
by chance. Both types of descendents contribute to the morphological averageness of the long-duration originator by pulling the
grand group average closer to the morphology of the originator
lineage. This is consistent with Wagner and Erwin’s (1995) finding that lineages that persist for long periods of time give rise
to more descendents. It also agrees with William’s desperation
hypothesis (1992, p. 132) which proposes that persistent forms
often generate numerous ephemeral forms. On the other hand, the
current discussion seems to contradict the finding that ancestral
species are statistically more likely to become extinct before their
coexisting descendents (Pearson 1998). However, rarer ancestral
species that persist despite their descendents will have truly long
EVOLUTION APRIL 2007
897
LEE HSIANG LIOW
durations, and hence are more relevant to the discussions here
regarding long-duration taxa in greater inclusive groups. Clearly,
more work is required to understand the links between diversification history and morphology.
Lineages with longer durations are significantly older. In
other words, they occur earlier in their group’s history or have
a greater geologic age, or time of first appearance. This pattern
holds even when only extinct groups are examined such that age
is not a constraint on observed durations. Lineages with longer
durations, even though they tend to be old, are not necessarily
morphologically similar to very basal forms (Liow 2004) or outgroups in a cladistic context. However they are often morphologically average when compared with lineages in the same inclusive clade through their collective history. These results suggest
that “successful” lineages, that is, those that lasted, have benefited by “being there early” and “looking right.” In contrast,
“unsuccessful” lineages, that is, those that did not last as long,
suffered from some combination of “not being there early” and
“not looking right.” That is to say, they appeared later and/or they
are experimental forms or morphologically “deviant” forms (see
also Williams 1992). This recalls niche preemption as discussed
by McKinney (1998, p. 7) wherein the first taxon that occupies
the empty “morphological niche” is not subsequently replaced
but morphologically emulated. Vermeij (1991) and Rosenzweig
and McCord (1991) discuss the idea of niche preemption in the
fossil record in terms of invasions and mass extinctions. I suggest that analogous processes of preemption can occur within a
clade during time intervals of background extinction, origination
and migration.
Morpho-duration distributions also have bearings on patterns
of morphospace occupation (see Foote 1997; Roy and Foote 1997
for reviews). Temporal trajectories of morphospace distribution
of individual lineages have received limited attention. Notably
though, it has been shown that as richness increases over time,
species are preferentially added to margins of empirical morphospace (Roy et al. 2001, references therein, also Ricklefs 2004).
The patterns shown in this study constrains this process such that
the addition of morphologically deviant taxa should not shift the
grand empirical morphological average of the inclusive clade directionally over time. If long-duration lineages are preferential
ancestors to many lineages in the inclusive clade, as suggested
in the previous paragraphs, then their descendents, even when
morphologically distant from their ultimate ancestor, should be
preferentially added to random parts of the margins of empirical
morphospace for the both patterns observed here and the findings
of Roy et al. (2001) to hold.
BIASES AND SOURCES OF ERROR
The only efficient way to widely investigate the relationship between morphology and duration is to sample the published lit-
898
EVOLUTION APRIL 2007
erature. However, there are a number of biases and sources of
error due to the usage of heterogeneous data collected for other
purposes, some of which have already been mentioned.
First, empirical values of grand average morphologies are
used as reference points but are calculated from incompletely sampled datasets. Hence they may not represent the true average morphology of the (para)clade in question. The completeness of studies could not be ascertained in a straightforward manner from the
original publications and hence were not considered here. However, if the morphological characters and the taxa are randomly
chosen with respect to lineage durations, the sample of taxa should
represent a spread of morphologies that allows a reliable estimate
of an average morphology of the group for their known history.
Moreover, the groups of datasets representing the three different
morpho-duration distributions have undifferentiable distributions
of the numbers of taxa and characters include in the analyses,
indicating that these factors are not systematically causing bias
in the results. Similarly, the characters coded may not adequately
represent the whole-organism morphology of the taxa (Table 1
reports cases where it was obvious only a small subset of morphological characters were used). Also, only adult morphologies are
reflected in the datasets used here.
Second, the taxa analyzed may not actually belong together
in a natural group due to incomplete knowledge of the (para)clade.
This problem is certainly not unique to this study and the fact that
the datasets originate from experts in their respective taxonomic
groups gives confidence that best efforts were made to include
only members that belong to the natural group in question. In a
similar vein, the taxonomic ranks of the taxa analyzed in each
dataset may not be equivalent. Although the identified cases are
few (Table 1), there could be cases in which the lineages are not
actually comparable even though they are hypothesized by the
original authors to represent the same ranks.
The ranks of stratigraphic ranges may not reflect the true
ranks of lineage durations. Rates of preservation can vary such
that taxa that in reality have their first or last appearances in
poorly preserved intervals will tend not to be sampled in them.
Thus they will have significantly shorter observed durations than
those taxa that first or last appear in better preserved time intervals and that range through poorly preserved intervals. However,
using inferred or raw durations provided largely similar results
of morpho-duration distributions for datasets where such inferences were available (16 of the 66 datasets), hence preservation
rates biases may not be large. One-sided range-truncations of extant taxa may similarly affect the true rank order of durations.
However, those datasets with extant representation do not show
a different distribution of results among the three scenarios of
morpho-duration distributions. Factors that promote high rates of
preservation in the fossil record, for example, wide geographic
ranges, numerical abundance, may cause relative overestimations
MORPHOLOGY AND LINEAGE DURATIONS
of the durations of some taxa. However, it has been shown that
even with sampling standardization that takes such high preservation rates into account, those taxa presumed to have long durations
remained classified as long-duration taxa (Liow 2007). In addition, morphological factors could be correlated with traits that
simultaneously influence preservation and lineage duration. Here
it is explicitly assumed that the lineages within datasets are similar enough morphologically such that preservation rates are not
likely to differ greatly. In addition, the type of depositional environment and the propensity for preservation as approximated
by ecological realms (terrestrial or aquatic) and phylogenetic representation played no systematic part in generating the resulting
morpho-duration patterns.
Conclusions
Many factors can influence the survivorship of individuals, populations and lineages during their lifetimes. These factors may
interact in a complex fashion so that it is difficult to tease apart
their individual contributions in holding populations and lineages
in prolonged stasis, in causing change, or accelerating extinction.
Despite this complexity, some factors have been clearly demonstrated to be important contributors to survivorship (Jablonski
2005) and others are beginning to be investigated as potential
properties that may confer greater lineage persistence in geologic
time. In particular, morphological distribution is predictably related to taxon duration. This relationship can be complicated by
taxonomic ranks, which have previously not been explicitly considered in discussions of persistent lineages. Contrary to the common idea that lineages with very long durations are special or
unique in some significant way, they often tend to be more average
than expected when compared with their relatives. This suggests
that deviations from locally optimal solutions, which evolutionary
processes have already found are usually not good candidates for
longer term survival.
ACKNOWLEDGMENTS
I thank S. Lidgard, L. Van Valen, P. Wagner, D. Jablonski, P. Harnik,
and C. Simpson for many helpful discussions on the topic of lineage
persistence and morphology. A. McGowan, three anonymous reviewers,
and the associate editor C. Janis made numerous suggestions that greatly
improved this manuscript. N. C. Stenseth and the folks at CEES supported
me with a conducive environment so this work could be completed. I also
thank T. F. Hansen who pointed out William’s desperation hypothesis
to me.
LITERATURE CITED
Adnet, S., and H. Cappetta. 2001. A palaeontological and phylogenetical analysis of squaliform sharks (Chondrichthyes : Squaliformes) based on dental characters. Lethaia 34:234–248.
Adrain, J. M., and G. D. Edgecomb. 1997. Silurian encrinurine trilobites from
the central Canadian Arctic. Palaeontol. Can. 14:1–109.
Adrain, J. M., and S. R. Westrop. 2001. Stratigraphy, phylogeny, and species
sampling in time and space. Pp. 29121 in J. M. Adrain, G. D. Edgecomb,
and B. S. Lieberman, eds. Fossils, phylogeny, & form: an analytical
approach. Kluwer/Plenum Press, New York.
Allmon, W. D. 1996. Evolution and systematics of Cenozoic American Turritellidae (Gastropoda). 1. Paleocene and Eocene species related to “Turritella mortoni Conrad” and “Turritella humerosa Conrad” from the U.S.
Gulf and Atlantic Coastal Plains. The Paleontological Research Institution, Ithaca, NY.
Alroy, J. 1995. Continuous track analysis—a new phylogenetic and biogeographic method. Syst. Biol. 44:152–178.
Alvarez, F., R. Jia-Yu, and A. J. Boucot. 1998. The classification of athyridid
brachiopods. J. Paleontol. 72:827–855.
Amati, L., and S. R. Westrop. 2004. A systematic revision of Thaleops (Trilobita: Illaenidae) with new species from the middle and late Ordovician
of Oklahoma and New York. J. Syst. Palaeontol. 2:207–256.
Anderson, L. C., and P. D. Roopnarine. 2003. Evolution and phylogenetic
relationships of Neogene Corbulidae (Bivalvia; Myoidea) of tropical
America. J. Paleontol. 77:1086–1102.
Angielczyk, K. D., and A. A. Kurkin. 2003. Phylogenetic analysis of Russian
Permian dicynodonts (Therapsida : Anomodontia): implications for Permian biostratigraphy and Pangaean biogeography. Zool. J. Linn. Soc.
139:157–212.
Anstey, R. L. 1978. Taxonomic survivorship and morphologic complexity in
Paleozoic Bryozoan genera. Paleobiology 4:407–418.
Baumiller, T. K. 1993. Survivorship analysis of Paleozoic Crinoidea—effect
of filter morphology on evolutionary rates. Paleobiology 19:304–321.
Beldade, P., K. Koops, and P. M. Brakefield. 2002. Developmental constraints
versus flexibility in morphological evolution. Nature 416:844–847.
Bloch, J. I., D. C. Fisher, K. D. Rose, and P. D. Gingerich. 2001. Stratocladistic analysis of Paleocene Carpolestidae (Mammalia, Plesiadapiformes)
with description of a new late Tiffanian genus. J. Vert. Paleont. 21:119–
131.
Bodenbender, B. E., and D. C. Fisher. 2001. Stratocladistic analysis of blastoid
phylogeny. J. Paleont. 75:351–369.
Boucot, A. J. 1975. Evolution and extinction rate controls. Elsevier Scientific
Publishing Company, Amsterdam, The Netherlands.
Boyajian, G., and T. Lutz. 1992. Evolution of biological complexity and its
relation to taxonomic longevity in the Ammonoidea. Geology 20:983–
986.
Brochu, C. A. 1997. Morphology, fossils, divergence timing, and the phylogenetic relationships of Gavialis. Syst. Biol. 46:479–522.
Brunet-Lecomte, P., and J. Chaline. 1991. Morphological evolution and
phylogenetic-relationships of the European ground voles (Arvicolidae,
Rodentia). Lethaia 24:47–53.
Buzas, M. A., and J. J. Culver. 1984. Species duration and evolution: benthic
Foraminifera on Atlantic continental margin of North America. Science
225:829–830.
Cairns, S. D. 2001. A generic revision and phylogenetic analysis of the Dendrophylliidae (Cnidaria: Scleractinia). Smithson. Contrib. Zool. 615:1–
88.
Caron, J. B., D. M. Rudkin, and S. Milliken. 2004. A new Late Silurian (Pridolian) naraoiid (Euarthropoda : Nektaspida) from the Bertie Formation of
southern Ontario, Canada-Delayed fallout from the Cambrian explosion.
J. Paleontol. 78:1138–1145.
Chen, Z.-Q., K. Kaihob, and A. D. George. 2005. Survival strategies of brachiopod faunas from the end-Permian mass extinction. Palaeogeog. Palaeoclimat. Palaeoecol. 224:232–269.
Damiani, R. J. 2001. A systematic revision and phylogenetic analysis of Triassic mastodonsauroids (Temnospondyli : Stereospondyli). Zool. J. Linn.
Soc. 133:379–482.
EVOLUTION APRIL 2007
899
LEE HSIANG LIOW
Dashzeveg, D., and J. Meng. 1998. New Eocene ctenodactyloid rodents from
the Eastern Gobi Desert of Mongolia and a phylogenetic analysis of
ctenodactyloids based on dental features. Am. Mus. Novit. 3246:1–20.
Dewing, K. 2004. Shell structure and its bearing on the phylogeny of Late Ordovicianarly Silurian strophomenoid brachiopods from Anticosti Island,
Quebec. J. Paleontol. 78:275–286.
Ebbestad, J. O. R., and G. E. Budd. 2002. Burlingiid trilobites from Norway, with a discussion of their affinities and relationships. Palaeontology
45:1171–1195.
Eldredge, N., J. N. Thompson, P. M. Brakefield, S. Gavrilets, D. Jablonski, J.
B. C. Jackson, R. E. Lenski, B. S. Lieberman, M. A. McPeek, and W.
Miller. 2005. The dynamics of evolutionary stasis. Paleobiology (Suppl.)
31:113–145.
Flessa, K. W., K. V. Powers, and J. L. Cisne. 1975. Specialization and evolutionary longevity in the Arthropoda. Paleobiology 1:71–91.
Foote, M. 1996. On the probability of ancestors in the fossil record. Paleobiology 22:141–151.
———. 1997. The evolution of morphological diversity. Annu. Rev. Ecol.
Syst. 28:129–152.
———. 1999. Morphological diversity in the evolutionary radiation of Paleozoic and post-Paleozoic crinoids. Paleobiology 25:1–115.
Forey, P. L. 1991. Latimeria chalumnae and its pedigree. Environ. Biol. Fishes
32:75–97.
Froehlich, D. J. 2002. Quo vadis eohippus? The systematics and taxonomy of
the early Eocene equids (Perissodactyla). Zool. J. Linn. Soc. 134:141–
256.
Gahn, F. J., and T. W. Kammer. 2002. The cladid crinoid Barycrinus from the
Burlington Limestone (early Osagean) and the phylogenetics of Mississippian botryocrinids. J. Paleontol. 76:123–133.
Gilinsky, N. L. 1988. Survivorship in the Bivalvia: comparing living and extinct genera and families. Paleobiology 14:370–386.
Gower, J. C. 1966. Some distance properties of latent root and vector methods
used in multivariate analysis. Biometrika 53:325–338.
Gradstein, F. M., J. G. Ogg, and A. G. Smith. 2004. A geologic time scale
2004. Cambridge Univ. Press, Cambridge, U.K., New York
Grande, L., and W. E. Bemis. 1998. A comprehensive phylogenetic study
of amiid fishes (Amiidae) based on comparative skeletal anatomy. An
empirical search for interconnected patterns of natural history. J. Vertebr.
Paleontol. 18:1–690.
Grether, G. F. 2005. Environmental change, phenotypic plasticity, and genetic
compensation. Am. Nat. 166:115–123.
Hansen, T. A. 1978. Larval dispersal and species longevity in lower Tertiary
gastropods. Science 199:885–887.
Hopkins, S. S. B. 2004. Phylogeny and biogeography of the genus Ansomys
Qiu, 1987 (Mammalia : Rodentia : Aplodontidae) and description of a
new species from the Barstovian (Mid-Miocene) of Montana. J. Paleontol. 78:731–740.
Jablonski, D. 1986a. Background and mass extinctions: the alternation of
macroevolutionary regimes. Science 231:129–133.
———. 1986b. Larval ecology and macroevolution in marine invertebrates.
Bull. Mar. Sci. 39:565–587.
———. 1994. Extinctions in the fossil record. Philos. Trans. R. Soc. Lond. B.
334:11–16.
———. 2005. Mass extinctions and macroevolution. Paleobiology 31:192–
210.
Jablonski, D., and D. M. Raup. 1995. Selectivity of end-Cretaceous marine
bivalve extinctions. Science 268:389–391.
Jablonski, D., and G. Hunt. 2006. Larval ecology, geographic range, and
species survivorship in Cretaceous mollusks: organismic versus specieslevel explanations. Am. Nat. 168:556–564.
Jackson, J. B. C. 1974. Biogeographic consequences of eurytopy and stenotopy
900
EVOLUTION APRIL 2007
among marine bivalves and their evolutionary significance. Am. Nat.
108:541–560.
Jeffery, C. H. 1998. Carrying on regardless: the echinoid genus Cyclaster at
the Cretaceous-Tertiary boundary. Lethaia 31:149–157.
Jeffery, C. H., and R. B. Emlet. 2003. Macroevolutionary consequences of
developmental mode in temnopleurid echinoids from the Tertiary of
southern Australia. Evolution 57:1031–1048.
Jernvall, J., and M. Fortelius. 2004. Maintenance of trophic structure in fossil
mammal communities: site occupancy and taxon resilience. Am. Nat.
164:614–624.
Kammer, T. W., T. K. Baumiller, and W. I. Ausich. 1997. Species longevity
as a function of niche breadth: evidence from fossil crinoids. Geology
25:219–222.
———. 1998. Evolutionary significance of differential species longevity
in Osagean-Meramecian (Mississippian) crinoid clades. Paleobiology
24:155–176.
Karasawa, H., and H. Kata. 2003. The family Goneplacidae MacLeay, 1838
(Crustacea: Decapoda: Brachyura): systematics, phylogeny, and fossil
records. Paleontol. Res. 7:129–151.
Koehl, M. A. R. 1996. When does morphology matter? Annu. Rev. Ecol. Syst.
27:501–542.
Lane, A., C. M. Janis, and J. J. Sepkoski. 2005. Estimating paleodiversities: a
test of the taxic and phylogenetic methods. Paleobiology 31:21–34.
Leighton, L. R., and C. G. Maples. 2002. Evaluating internal versus external
characters: Phylogenetic analyses of the Echinoconchidae, Buxtoniinae,
and Juresaniinae (phylum Brachiopoda). J. Paleontol. 76:659–671.
Levinton, J. S., and L. Ginzburg. 1984. Repeatability of taxon longevity in
successive Foraminifera radiations and a theory of random appearance
and extinction. Proc. Natl. Acad. Sci. USA 81:5478–5481.
Liow, L. H. 2004. A test of Simpson’s “Rule of the survival of the relatively
unspecialized” using fossil crinoids. Am. Nat. 164:431–443.
———. 2006. Do deviants live longer? Morphology and longevity in trachyleberidid ostracodes. Paleobiology 32:55–69.
———. 2007. Does versatility as measured by geographic range, bathymetric
range and morphological variability contribute to taxon longevity? Glob
Ecol Biogeogr 16:117–128.
McGowan, A. J. Does shape matter? Morphological diversity and differential
survivorship among Triassic ammonoid genera. Hist. Biol. In press.
McKinney, M. L. 1998. Biodiversity dynamics: niche preemption and saturation in diversity equilibria. Pp. 1–15 in M. L. McKinney and J. A.
Drake, eds. Biodiversity dynamics: turnover of populations, taxa, and
communities. Columbia Univ. Press, New York.
Merilä, J., B. C. Sheldon, and L. E. B. Kruuk. 2001. Explaining stasis: microevolutionary studies in natural populations. Genetica 112–113:199–
222.
Michaux, B. 1989. Cladograms can reconstruct phylogenies—an example
from the fossil record. Alcheringa 13:21–36.
Monks, N. 1999. Cladistic analysis of Albian heteromorph ammonites.
Palaeontology 42:907–925.
Monks, N. 2002. Cladistic analysis of a problematic ammonite group: The
Hamitidae (Cretaceous, Albian-Turonian) and proposals for new cladistic terms. Palaeontology 45:689–707.
Monks, N., and E. Owen. 2000. Phylogeny of Orbirhynchia pettitt, 1954 (Brachiopoda : Rhynchonellida). Palaeontology 43:871–880.
Nutzel, A., D. H. Erwin, and R. H. Mapes. 2000. Identity and phylogeny of
the late Paleozoic Subulitoidea (Gastropoda). J. Paleontol. 74:575–598.
Oji, T. 1996. Is predation intensity reduced with increasing depth? Evidence
from the west Atlantic stalked crinoid Endoxocrinus parrae (Gervais)
and implications for the Mesozoic marine revolution. Paleobiology
22:339–351.
O’Keefe, F. R. 2004. Preliminary description and phylogenetic position of a
MORPHOLOGY AND LINEAGE DURATIONS
new Plesiosaur (Reptilia : Sauropterygia) from the Toarcian of Holzmaden, Germany. J. Paleontol. 78:973–988.
Pearson, P. N. 1998. Speciation and extinction asymmetries in paleontological
phylogenies: evidence for evolutionary progress? Paleobiology 24:305–
335.
Popov, L. E., I. F. Nikitin, and E. V. Sokiran. 1999. The earliest atrypides and
athyridides (brachiopoda) from the Ordovician of Kazakhstan. Palaeontology 42:625–661.
Raup, D. M. 1985. Mathematical models of cladogenesis. Paleobiology 11:42–
52.
Ricklefs, R. E. 2004. Cladogenesis and morphological diversification in
passerine birds. Nature 430:338–341.
Roopnarine, P. D. 2001. A history of diversification, extinction, and invasion in
tropical America as derived from species-level phylogenies of chionine
genera (Family Veneridae) (Reprinted). J. Paleontol. 75:644–657.
Roy, K., and M. Foote. 1997. Morphological approaches to measuring biodiversity. TREE 12:277–281.
Roy, K., D. P. Balch, and M. E. Hellberg. 2001. Spatial patterns of morphological diversity across the Indo-Pacific: analyses using strombid gastropods.
Proc. R. Soc. Lond. B. 268:2503–2508.
Rosenzweig, M. L., and R. D. McCord. 1991. Incumbent replacement—
evidence for long-term evolutionary progress. Paleobiology 17:202–213.
Rutherford, S. L. 2000. From genotype to phenotype: buffering mechanisms
and the storage of genetic information. BioEssays 22:1095–1105.
Schopf, J. W. 1994. Disparate rates, differing fates: tempo and mode of evolution changed from the Precambrian to the Phanerozoic. Proc. Natl. Acad.
Sci. USA 91:6735–6742.
Schneider, J. A. 1995. Phylogeny of the Cardiidae (Mollusca, Bivalvia): Protocardiinae, Laevicardiinae, Lahilliinae, Tulongocardiinae subfam. n. and
Pleuriocardiinae subfam. n. Zool. Scr. 24:321–346.
Schwenk, K., and G. P. Wagner. 2001. Function and the evolution of phenotypic
stability: connecting pattern to process. Am. Zool. 41:552–563.
Simpson, G. G. 1944. Tempo and mode in evolution. Columbia Univ. Press,
New York.
———. 1953. The major features of evolution. Columbia Univ. Press, New
York.
Schneider, J. A. 1995. Phylogeny of the Cardiidae (Mollusca, Bivalvia): Protocardiinae, Laevicardiinae, Lahilliinae, Tulongocardiinae subfam. n. and
Pleuriocardiinae subfam. n. Zool. Scr. 24:321–346.
Smith, A. B. 1988. Patterns of diversification and extinction in Early Paleozoic
echinoderms. Palaeontology 31:799–828.
Smith, A. B., and M. A. Arbizu. 1987. Inverse larval development in a Devonian
Edrioasteroid from Spain and the phylogeny of Agelacrinitinae. Lethaia
20:49–62.
Smith, A. B., G. L. J. Paterson, and B. Lafay. 1995. Ophiuroid phylogeny and
higher taxonomyorphological, molecular and paleontological perspectives. Zool. J. Linn. Soc. 114:213–243.
Smith, A. B., and C. W. Wright. 1993. British Cretaceous echinoids. Part 3,
Stirodonta 2 (Hemicidaroida, Arbacioida and Phymosomatoida, Part 1).
Monogr. Palaeontol. Soc. 593:199–267.
Stanley, S. M. 1979. Macroevlution. The John Hopkins Univ. Press, Baltimore
and London.
Tinn, O., and T. Meidla. 2004. Phylogenetic relationships of Early-middle
Ordovician ostracods of Baltoscandia. Palaeontology 47:199–221.
Van Valen, L. 1982. Integration of species: stasis and biogeography. Evol.
Theory 6:99–112.
Vermeij, G. J. 1991. Anatomy of an invasion —the trans-Arctic interchange.
Paleobiology 17:281–307.
Vermeij, G. J., and S. J. Carlson. 2000. The muricid gastropod subfamily
Rapaninae: phylogeny and ecological history. Paleobiology 26:19–46.
Wackerly, D. D., W. Mendenhall III, and R. L. Scheaffer. 2002. Mathematical
statistics with applications. Duxbury.
Wagner, P. J. 1997. Patterns of morphologic diversification among the Rostroconchia. Paleobiology 23:115–150.
———. 1999. The utility of fossil data in phylogenetic analyses: a likelihood example using Ordovician-Silurian species of the Lophospiridae
(Gastropoda : Murchisoniina). Am. Malacol. Bull. 15:1–31.
———. 2000. Exhaustion of morphologic character states among fossil taxa. Evolution 54:365–386. Available from http://pjw3.fmnh.
org/EvolutionMatrices˙2000.html. Accessed jan. 2006.
Wagner, P. J., and D. H. Erwin. 1995. Phylogenetic patterns as tests of speciation models. Pp. 87–122 in D. H. Erwin and R. L. Anstey, eds. New
Approaches to Speciation in the Fossil Record. Columbia Univ. Press,
New York.
Wainwright, P. C., and S. M. Reilly. 1994. Ecological morphology: integrative
organismal biology. Univ. Chicago Press, Chicago.
Wake, D. B., G. Roth, and M. H. Wake. 1983. On the problem of stasis in
organismal evolution. J. Theo. Biol. 101:211–224.
Ward, P. D., and P. W. Signor. 1983. Evolutionary tempo in Jurassic and
Cretaceous ammonites. Paleobiology 9:183–198.
Williams, G. C. 1992. Natural selection: domains, levels and challenges. Oxford Univ. Press, New York and Oxford.
Wills, M. A. 2001. Morphological disparity : a Primer. Pp. 55–144 in J. M.
Adrain, G. D. Edgecombe and B. S. Lieberman, eds. Fossils, phylogeny,
and form: an analytical approach. Kluwer Academic and Plenum Publishers, New York.
Yates, A. M., and A. A. Warren. 2000. The phylogeny of the ’higher’ temnospondyls (Vertebrata : Choanata) and its implications for the monophyly and origins of the Stereospondyli. Zool. J. Linn. Soc. 128:77–121.
Associate Editor: C. Janis
Supplementary Material
The following supplementary material is available for this article:
Appendices S1–S4.
This material is available as part of the online article from:
http://www.blackwell-synergy.com/doi/abs/10.1111/j.1558-5646.2007.00077.x
(This link will take you to the article abstract).
Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by
the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
EVOLUTION APRIL 2007
901