Paleobiology, 31(2), 2005, pp. 133 145
The dynamics of evolutionary stasis
Niles Eldredge, John N. Thompson, Paul M. Brakefield, Sergey Gavrilets,
David Jablonski, Jeremy B. C. Jackson, Richard E. Lenski, Bruce S. Lieberman,
Mark A. McPeek, and William Miller III
Abstract. The fossil record displays remarkable stasis in many species over long time periods, yet
studies of extant populations often reveal rapid phenotypic evolution and genetic differentiation
among populations. Recent advances in our understanding of the fossil record and in population
genetics and evolutionary ecology point to the complex geographic structure of species being fun-
damental to resolution of how taxa can commonly exhibit both short-term evolutionary dynamics
and long-term stasis.
Niles Eldredge. Division of Paleontology, American Museum of Natural History, Central Park West at Sev-
enty-ninth Street, New York, New York 10024. E-mail: [e-mail]
John N. Thompson. Department of Ecology and Evolutionary Biology, A316 Earth and Marine Sciences Build-
ing, University of California, Santa Cruz, California 95060. E-mail: [e-mail]
Paul M. Brakefield. Institute of Biology, Leiden University, Post Office Box 9516, 2300 RA Leiden, The
Netherlands. E-mail: [e-mail]
Sergey Gavrilets. Department of Ecology and Evolutionary Biology and Department of Mathematics, Uni-
versity of Tennessee, Knoxville, Tennessee 37996. E-mail: [e-mail]
David Jablonski. Department of Geophysical Sciences, 5734 South Ellis Avenue, University of Chicago,
Chicago, Illinois 60637. E-mail: [e-mail]
Jeremy B. C. Jackson. Scripps Institution of Oceanography, University of California, San Diego, La Jolla,
California 92039. E-mail: [e-mail]
Richard E. Lenski. Center for Microbial Ecology, Michigan State University, East Lansing, Michigan
48824. E-mail: [e-mail]
Bruce S. Lieberman. Departments of Geology and Ecology and Evolutionary Biology, University of Kansas,
120 Lindley Hall, Lawrence, Kansas 66045. E-mail: [e-mail]
Mark A. McPeek. Department of Biological Sciences, Dartmouth College, Hanover, New Hampshire 03755.
E-mail: [e-mail]
William Miller III. Department of Geology, Humboldt State University, 1 Harpst Street, Arcata, California
95521. E-mail: [e-mail]
Accepted: 17 April 2004
Introduction have been adopted and results obtained in all
these fields. Our basic conclusion that stasis
The pronounced morphological stability
derives from the geographic structure and
displayed by many fossil species (Eldredge
partitioning of genetic information within
1971; Eldredge and Gould 1972; Gould and
widespread species is derived from a con-
Eldredge 1977; Stanley and Yang 1987; Jackson
sideration of all known population genetic
and Cheetham 1999; Jablonski 2000), often for
processes that promote (or conversely hinder)
millions of years, contrasts sharply with the
genetic change, as well as from analysis of
rapid, often adaptive, evolutionary changes
data from the fossil record.
documented in many extant species (Reznick
Stasis is generally defined as little or no net
et al. 1997; Thompson 1998; Huey et al. 2000;
accrued species-wide morphological change
Thomas et al. 2001). If evolutionary change oc-
during a species-lineage s existence up to mil-
curs frequently within populations, why is it
lions of years instantly begging the question
that in some species so little of it is conserved
of the precise meaning of   little or no  net
and translated through time as net change? In
evolutionary change. All well-analyzed fossil
this paper we examine what paleobiologists,
species lineages, as would be expected, dis-
population geneticists, and evolutionary ecol- play variation within and among populations,
ogists have learned about stasis and rapid evo- but the distribution of this variation typically
lution over the past decade as new approaches remains much the same even in samples sep-
2005 The Paleontological Society. All rights reserved. 0094-8373/05/3102-0010/$1.00
134 NILES ELDREDGE ET AL.
FIGURE 1. Species stasis in the face of ongoing population-level evolution. Species (lineages 1, 2, and 3 on the left)
exhibit negligible net phenotypic changes, while their component population systems (on the right) continually
differentiate, fuse, or go extinct. Stasis is occasionally broken by establishment and spread of novel phenotypes (s);
when this is matched with ecological opportunity, highly differentiated new lineages (sm) may be formed that
eventually develop internal (population) dynamics and geographic structure resulting, again, in stasis. (In this view,
species-lineages consist of anastomosing population systems and, at the same time, belong to clades composed of
similar lineages).
arated by millions of years (Fig. 1). This view 1987) and bryozoans (Jackson and Cheetham
of fossil variation has been reinforced over the 1999). Inventories of evolutionary tempo and
past decade as paleontological studies have mode across entire clades are sparse, but Jack-
applied higher sampling intensities in time son and Cheetham s (1999) survey of well-
and space, improvements in both relative and documented case studies in the Neogene fossil
absolute stratigraphic dating, more compre- record found 52 instances of stasis and only
hensive use of multivariate statistical analysis, two instances of anagenesis in nine benthic
and better controls for sampling biases. macroinvertebrate clades, and eight instances
Although it is now clear that some fossil of stasis as opposed to 10 12 instances of ana-
species lineages do indeed accrue morpholog- genesis in marine microplankton. Anagenesis
ical change through time (Geary 1995), it is occurs in only eight of 88 trilobite lineages in
also now evident that many do not. Well-doc- the Ordovician of Spitsbergen, and in but one
umented examples of stasis range from Paleo- of 34 scallop lineages in the northern Euro-
zoic brachiopods (Lieberman et al. 1995) to pean Jurassic (Jablonski 2000).
late Cenozoic bivalves (Stanley and Yang Studies of extant taxa with rich fossil rec-
DYNAMICS OF EVOLUTIONARY STASIS 135
ords provide mounting evidence that morpho- punctuated equilibria (Eldredge 1971; Eld-
logically defined species-level lineages recog- redge and Gould 1972). Recently, Webster,
nized in fossil sequences often correspond to Payne, and Pagel (2003), in their analysis of
genetically defined species in the modern bi- speciation events and underlying genetic
ota (Jablonski 2000). Such studies are crucial
change in 56 phylogenies, concluded that
to the demonstration that patterns of stasis in
  rapid genetic evolution frequently attends
the fossil record constitute a genuine problem
speciation,  and that their results provide a
for evolutionary theory. Perhaps the most rig-
  genetic component  to the pattern of stasis
orous and detailed of such studies (and one
and change of morphological traits seen so
that has proven compelling to population ge-
commonly in the fossil record.
neticists) are those on tropical American Neo-
Our purpose here is to explore further the
gene cheilostome bryozoans (Jackson and
dynamics generating such patterns, particu-
Cheetham 1999). Cheilostomes are small,
larly insofar as stasis itself is concerned. Given
clonal marine animals that grow in plantlike
patterns of changes in heritable phenotypic
shapes by budding modules (zooids) to form
variation and genetic variation commonly
a colony. They are abundant in Recent seas
seen in local populations, what factors prevent
and in the fossil record. In the tropical Amer-
such change from becoming species-wide? Do
ican genera Metrarabdotos and Stylopoma, all
novelties arise only at speciation, or do they
long-ranging species (11 in each genus) per-
arise but are typically not conserved through-
sisted essentially morphologically unchanged
out the history of species perhaps further
for 2 16 Myr. New species appear abruptly in
suggesting that speciation conserves rather
the fossil record, with morphological change
than prompts the generation of novelty? Pre-
occurring within the limits of stratigraphic
vious authors (Darwin 1871; Ohta 1972; Fu-
resolution of sampling (approximately
tuyma 1987; Eldredge 1989; Lieberman et al.
150,000 years). Studies of extant species in
1995) have discussed the difficulties inherent
these genera indicate that morphological sta-
in conserving evolutionary novelties arising in
sis also reflects stasis in key life history traits,
local populations and their spread over the en-
with occasional rapid change. For example,
tire range of a far-flung, heterogeneous spe-
the size of larval brood chambers, which is
cies. Futuyma (1987) in particular has dis-
correlated with larval size, differs by up to
cussed the closely related corollary that spe-
twofold among closely related species, and en-
ciation may be the key to the phylogenetic
tirely arborescent species have given rise to
conservation of such novelties. More recently,
entirely encrusting species. Such examples
the geographic mosaic of ongoing local ad-
show that stasis can include reproductive and
aptation has become the very foundation for
behavioral characteristics in addition to pure
new views of how coevolving interactions be-
morphology. We find this example and other
tween species persist over long periods of time
such case studies compelling evidence that
in a constantly changing world (Thompson
morphological stasis is a common pattern in
1994, 1999a,b).
the fossil record, which thus requires an ex-
What, then, constrains the species-wide
amination of how evolutionary and ecological
spread of evolutionary change when experi-
processes can account for it.
mental and field data clearly show that the po-
If many, perhaps even most, species accrue
tential for rapid change within populations is
little morphological change during their life-
nearly always present? We divide the question
times, then a corollary is immediately raised:
into three stages related to the establishment
the possibility that much of the morphological
of evolutionary change in a geographically
change accrued within evolutionary lineages
over time is concentrated in relatively brief ep- heterogeneous world: origin, local population
establishment, and species-wide spread. Our
isodes of speciation. Mayr (1954) suggested a
analysis of studies from the past decade, in-
link between speciation and evolutionary
change, and stasis morphological change cluding examples drawn from our own work,
concentrated at speciation events is the core of suggests that patterns and processes related to
136 NILES ELDREDGE ET AL.
geographic structure contribute importantly variation is present, however, evolutionary po-
to the maintenance of stasis. tential is not equal in all traits, and the origi-
nation of useful novelties may depend upon
Origin, Local Population Establishment,
mutations appearing in a particular sequence
and Species-wide Spread
(Mani and Clarke 1990). Antagonistic pleiot-
To be preserved in the fossil record with any ropy (leading to negative genetic correla-
reasonable likelihood, a novel genotype must tions), epistasis, and linkage disequilibrium
originate, become established in a local pop- can all constrain the generation of novel ge-
ulation, and then spread and increase in num- notypes, even when standing genetic variation
bers across a large geographic area. Failure to is not limited by population size or the pre-
complete all three of these stages will result in vious history of selection (Barton and Par-
stasis in the fossil record. Consequently, if we tridge 2000). Some artificial-selection experi-
are to understand the evolutionary dynamics ments have shown that rates of phenotypic
of stasis, we need to understand where most change may decelerate during prolonged di-
failures occur along this sequence of origin rectional selection (Falconer and Mackay
and spread of novelty. Many earlier attempts 1996). This pattern has often been attributed
to reconcile our understanding of the evolu- to the depletion of the genetic variation for the
tionary dynamics of extant species with the selected trait that was present in the founding
paleobiological evidence for stasis focused on population or, alternatively, depletion of var-
the role of genetic constraints and stabilizing iation in fitness more generally, such as when
selection in preventing the origin and estab- selection on some other aspect of organismal
lishment of novelty within local populations performance opposes the response to artificial
(Charlesworth et al. 1982; Van Valen 1982; selection (Barton and Partridge 2000; Falconer
Levinton 1983; Maynard Smith 1983; Wake et and Mackay 1996). Consistent with these ex-
al. 1983; Williamson 1987). More recent math- planations, response to selection can be accel-
ematical and empirical studies have refined erated by increasing population size, which
our understanding of the roles of these evo- both increases the overall level of genetic var-
lutionary forces, and they have shown that the iation and opens new permissible directions
spatial structure of species strongly influences (  ridges  ) available to selection in the multi-
the pattern of establishment of novel types. dimensional adaptive landscape (Weber
Constraints on the Origin and Local Establish- 1996).
ment of Novelty. From a theoretical perspec- Pronounced decelerations in rates of phe-
tive, the origin of novel genotypes involves a notypic evolution have also been observed
set of processes (mutation and recombination) over thousands of generations in asexual pop-
distinct from those processes that determine ulations of Escherichia coli founded from a sin-
the local fate of the variants that are produced gle cell (Cooper and Lenski 2000). In these
(drift and selection). From an empirical per- populations, new mutations provide the only
spective, however, the actual rate of produc- source of genetic variation, and this mutation-
tion of novel variants is very rarely observed al source continues indefinitely. In this case,
directly. Instead, the failure to produce nov- stasis cannot derive from depletion of preex-
elty, on the one hand, versus the failure of nov- isting variation, nor from exhaustion of genet-
elties to become locally established, on the ic variation more generally. In fact, the
other hand, must often be inferred indirectly amount of genetic variation increased in these
from the dynamics of experimental systems. populations even as the rate of phenotypic
Therefore, we combine our analysis of these evolution declined (Sniegowski et al. 1997). In-
two dynamical stages in the section that fol- stead, these populations evidently ap-
lows. proached a local adaptive peak or plateau, at
The simplest potential explanations for sta- which point most potential (i.e., genetically
sis are exhaustion of standing genetic varia- accessible) beneficial mutations were fixed.
tion or the limited production of useful nov- Consistent with this explanation, the rate of
elties within populations. Even when genetic adaptive evolution was re-accelerated by per-
DYNAMICS OF EVOLUTIONARY STASIS 137
turbing populations from their proximity to ulations (Wade and Goodnight 1998). Envi-
an adaptive peak, either by changing the en- ronmental stress may disrupt developmental
vironment (Travisano et al. 1995) or by intro- stability sufficiently to uncover latent genetic
ducing deleterious mutations (Moore et al. variance that can promote evolvability (Ruth-
2000). erford and Lindquist 1998). Yet other process-
These studies show that relative stasis can es including gene or genome duplication,
arise fairly quickly following periods of rapid polyploidy, hybridization, and horizontal
adaptive evolution. They also indicate that the gene transfer can further promote novel
exhaustion of beneficial variants whether paths of evolution (Rieseberg 1997; Soltis and
preexisting or potentially accessible by muta- Soltis 1999; Lynch and Force 2000; Sandstrom
tion can contribute to stasis. However, the et al. 2001). Consequently, it increasingly
depletion of standing variation is relevant seems that neither an absence of genetic vari-
only in small populations, which contribute ation nor genetic constraints are sufficient to
very little to the fossil record. Species-wide account for long-term stasis.
depletion of accessible beneficial mutations re- Expression of advantageous genetic varia-
quires a degree of environmental constancy tion in highly variable environments, howev-
that is not typical of the earth s history (Lam- er, may constrain the breaking of stasis within
beck and Chappell 2001; Zachos et al. 2001). local populations. Recent theoretical studies
More likely, genetic and developmental cor- of multidimensional genotype space have
relations among traits can also influence both demonstrated the possibility of prolonged
the direction and extent of change in local phenotypic change within local populations
populations, and advances in evolutionary de- by a chain of substitutions that are nearly neu-
velopmental biology are suggesting the extent tral with respect to overall fitness in the ab-
to which these genetic interactions may influ- sence of a highly variable environment (Gav-
ence stasis. For example, some experiments on rilets 1997). Only a small proportion of mu-
butterfly wing patterns show that multiple tations with significant phenotypic effects are
eyespots are made by the same developmental expected to be advantageous or even neutral.
pathway and, consequently, there exist strong The more variable the environment over time,
genetic correlations among them. Selection to the more restricted the range of these geno-
increase the size of the posterior eyespot on types with equal or higher fitness, because
the forewing of Bicyclus anynana, in the ab- each genotype must function under a wide
sence of any selection on the anterior eyespot range of environmental conditions.
(Beldade et al. 2002), will typically increase When stasis breaks down, it may do so ei-
the size of both eyespots. Nonetheless, selec- ther in large or in small populations. Consid-
tion can readily uncouple the two eyespots to ering both the production of mutations and
produce highly divergent morphologies in all their subsequent fate, advantageous mutations
directions of morphological space (Fig. 2). In- will become established more often in larger
deed, novel patterns not seen in any related than in smaller populations. An environmen-
species can be obtained after 25 or so gener- tal change, by redefining the   optimum  phe-
ations. These results indicate that genetic and notype, may result in increasing the probabil-
developmental processes can produce genetic ity of mutations being conditionally advanta-
correlations and favor evolution along paths of geous or neutral, thereby promoting evolu-
least resistance, but they need not absolutely tionary change. On the other hand, decreasing
constrain the process of adaptive radiation population size will increase the role of sto-
(Brakefield et al. 2003). chastic fluctuations, creating an opportunity
In fact, recent studies have revealed a vari- to overcome stabilizing selection or incum-
ety of genetic mechanisms that may overcome bency effects (Barton and Charlesworth 1984)
constraints imposed by gene interaction. Epi- and facilitating evolution along an adaptive
static components of genetic variance may be ridge of genotypes that are nearly equal in fit-
converted into additive variance, promoting ness (Gavrilets 1999). Strong competition for a
evolutionary change in small, perturbed pop- resource may potentially lead to sympatric or-
138 NILES ELDREDGE ET AL.
FIGURE 2. Analysis of a potential evolutionary constraint. A, Occupation by species of the butterfly genus Bicyclus
of morphological space for the pattern of the forewing eyespot size. Names of representatives from among the 80
or so species are given. B, Responses obtained over 25 generations of artificial selection in replicate lines of B. an-
ynana. Results show that butterflies similar to each corner pattern were produced from standing genetic variation
in a single laboratory stock, including one morphology not seen in any extant species. Crosses indicate butterflies
from the base population, and open symbols show samples from generation 25 in each direction of selection (green
arrow) together with a representative forewing. Redrawn from Beldade et al. 2002.
DYNAMICS OF EVOLUTIONARY STASIS 139
igin and within-deme establishment of genet- fish or large dragonflies as the top predators
ic novelties (Kawata 2002). More complete ces- (McPeek and Brown 2000). Enallagma species
sation of gene flow can result in rapid evolu- differ in their vulnerability to these predators
tionary change in a population experiencing a and are thus capable of living with only one
novel environment (Garcia-Ramos and Kirk- of them. Species that coexist with fish use
patrick 1997). crypsis to avoid predators, whereas species
These theoretical expectations on the con- that coexist with large dragonflies are more
ditions allowing occasional breakdown of sta- active and swim away from attacking preda-
sis receive support from experimental studies tors (McPeek 1998). Moreover, several Enallag-
during the past decade. Besides the decelera- ma species are found in each lake type, and co-
tion in phenotypic evolution found during the occurring species are phenotypically very
long-term experiments in E. coli (Lenski and similar (McPeek 2000). Lakes with fish are the
Travisano 1994), both performance and mor- ancestral habitat for the genus, and at least
phology show a stair-step dynamic over two independent invasions by damselflies into
shorter periods. Most of the changes in the the dragonfly lake environment have occurred
first 3000 generations were concentrated in a (McPeek and Brown 2000). These habitat
few episodes that appeared instantaneous at a shifts have been accompanied by rapid evo-
100-generation sampling interval. These epi- lution in a number of morphological, behav-
sodes have a simple explanation: each step in ioral, and biochemical characters that enhance
performance reflects a selective fixation of a burst swimming speed because of selection
beneficial mutation, and the morphological imposed by dragonfly predators in the new
changes are pleiotropic effects of these muta- environment (McPeek 2000; McPeek and
tions. Brown 2000). It may have taken the invading
Rapid diversification of a lineage may there- lineages only a few hundred years to gain a
fore often involve the invasion of a new selec- high degree of local adaptation to their new
tive environment by one or a few local popu- environment (McPeek 1997).
lations. The breakdown of stasis occurs as a lo- Such rapid evolutionary change would ap-
cal population adapts rapidly to an initially pear saltatory in the fossil record. In contrast,
inhospitable habitat before it would otherwise rates of evolution in these characters within
be driven extinct (Gomulkiewicz and Holt the fish lake environment are very slow. Mil-
1995). Rapid, pulsed diversification of some lions of years of evolution within the fish lake
phytophagous insects as they colonize new environment have produced few or no differ-
host taxa in local populations has long been a ences among species in many other characters
working model for studies of plant-insect in- that are important in determining their eco-
teractions (Ehrlich and Raven 1964). Molecu- logical performance (McPeek 2000; McPeek
lar phylogenetic analyses of insect taxa during and Brown 2000). Importantly, shifts to drag-
the past decade have provided evidence for onfly lakes and accompanying rapid evolution
such bouts of rapid diversification at the bases have been rare events, occurring in only one
of clades, as species colonize new host line- of the two primary clades of Enallagma. That
ages (Pellmyr et al. 1998). Similarly, the occa- clade has a number of phenotypic characters
sional invasion by E. coli of thermally stressful that are already similar to phenotypes favored
environments, beyond the tolerance limits of by selection in dragonfly lakes (McPeek 2000).
ancestral populations, fits this model (Mon- Hence, there appears to be a fundamental
gold et al. 2001). niche conservatism that dooms shifts by most
One well-studied example of invasion of a populations to failure, thereby contributing to
novel environment leading to the breakdown stasis within many Enallagma.
of stasis and the generation of evolutionary Recent studies therefore suggest that the ab-
novelty correlated with speciation is found sence of useful novelties, or their failure to be-
in the diversification of damselflies. Enallagma come established within local populations,
damselflies have diversified in North America may contribute to stasis in certain limited cas-
into permanent ponds and lakes with either es. But more generally, the field of evolution-
140 NILES ELDREDGE ET AL.
ary ecology has clearly shown the ability of lo- trinsic advantage over invaders because they
cal populations to evolve rapidly under occur at high relative frequency or density
changing conditions. Consequently, species- (Gomulkiewicz and Holt 1995; McPeek 2000).
wide stasis would seem to require additional The incumbency of established species can be
constraints acting above the level of local pop- further maintained through effects on hy-
ulations. brids. Recent mathematical models show that
Species-wide Spread. A key change in pop- if hybrids between the novel and incumbent
ulation genetic theory and evolutionary ecol- forms have reduced fitness, then the chance of
ogy over the past decade has been the increas- spread of the novel form is further reduced
ing incorporation of geographic structure into (Gavrilets 1996; Coyne et al. 1997). Moreover,
our understanding of the evolutionary dy- through asymmetric gene flow most hybrids
namics of species. We now know from various are likely to occur within the population
genetic modeling approaches that spatial where the novel genotype originated, because
structure can decrease the likelihood of re- the absolute numbers will often be less than
gional extinction, maintain genetic polymor- surrounding populations. This asymmetric
phisms across populations, and shape evolu- gene flow will therefore minimize the chance
tionary and coevolutionary trajectories (Gan- that a novel form will rise to high frequencies
don et al. 1996; Thrall and Burdon 1997; Go- elsewhere.
mulkiewicz et al. 2000; Nuismer et al. 2000). At The development of metapopulation theory
the same time, a burgeoning number of stud- (Hanski and Gilpin 1997) has provided yet ad-
ies in molecular ecology and evolutionary ditional insights into the problem of spread in
ecology reveal even more widespread genetic novel forms (Lande 1985; Tachida and Ilizuka
differentiation among populations than was 1991). Some current models suggest that high
apparent from earlier studies that often un- population turnover rates can reduce the
derestimated spatial genetic structure. These chances of establishment and spread of novel
modeling and empirical results together sug- genotypes, unless those genotypes are fa-
gest that the geographic genetic structure of vored by their very rarity through negative
species must be a central component of any frequency-dependent selection as occurs in
overall theory resolving the discrepancy be- gene-for-gene coevolution between some
tween short-term dynamics and long-term plants and pathogens (Burdon and Thrall
stasis. 1999; Gandon et al. 1996). This kind of nega-
Novel forms must spread beyond their site tive frequency-dependent selection, which
of origin if they are to have a reasonable maintains polymorphisms by favoring rare
chance of being preserved in the fossil record. genotypes within and among populations,
If a local population is already reproductively may also maintain stasis within a species rath-
isolated from its neighbors, then novel forms er than lead to diversification. When meta-
must successfully expand beyond their initial population structure is coupled with hetero-
geographic limits. Alternatively, if the local geneous selection across landscapes, it may
population is still genetically connected to become even more difficult for novel geno-
other populations, then a novel form must be types to spread.
able to spread across those other populations Paleobiologists have argued that wide-
if it is to become sufficiently widespread to spread species are expected to exhibit slower
leave a record of the change. In both cases the rates of species-wide evolution than species
key problem is expansion of geographic range with small ranges, because natural selection
(Kirkpatrick and Barton 1997; Thomas et al. will not be consistently directional across
2001). We now know that, even once estab- space and time (Eldredge 2003; Jablonski
lished locally, novel forms may face large hur- 2000; Lieberman et al. 1995) e.g., ecological
dles in spreading beyond their site of origin. conditions acting on local populations of
These spatially induced hurdles may be the American robins in the southwestern United
most potent evolutionary forces maintaining States are clearly different from those present
stasis. Established species often have an in- in the deep woodlands of the Northeast. That
DYNAMICS OF EVOLUTIONARY STASIS 141
overall expectation is supported by popula- edly important to the ecological dynamics of
tion genetic theory, which suggests that it is species. Moreover, it may be crucial for keep-
difficult for a mutant to be advantageous un- ing coevolving species in the evolutionary
der all conditions required by a highly hetero- game as one species or the other temporarily
geneous environment (Ohta 1972). Consistent gains the upper hand in different environ-
with the expectation that most mutations that ments. But most of these dynamics may not
are locally adaptive would not be globally ad- result in much net change at the species level.
vantageous, lines of E. coli adapted to a glu- Recent mathematical models have indicated
cose-containing environment for 20,000 gen- that geographically structured coevolution
erations tend to have reduced performance on can actually constrain the escalation of antag-
a range of other substrates (Cooper and Len- onistic arms races. These interactions may
ski 2000). continually recycle defenses and counterde-
The developing mathematical theory of spe- fenses through frequency-dependent selec-
cies ranges provides additional indications tion, because geographic structure may main-
that the spatial structure of habitats and het- tain the polymorphisms on which frequency-
erogeneous selection may be important sourc- dependent selection depends (Gandon et al.
es of stasis. Gene flow from the center of a spe- 1996; Gomulkiewicz et al. 2000; Nuismer et al.
cies range can impede novel adaptation at the 2000). Long-term studies of gene-for-gene co-
periphery and prevent the range from ex- evolution within natural populations support
panding outward (Kirkpatrick and Barton this mathematical prediction (Burdon and
1997). The problems of spatial structure and Thrall 2000). Similarly, geographic structure
heterogeneous selection may therefore con- may stabilize some kinds of coevolved mutu-
tribute to the kind of sustained habitat track- alisms, by maintaining previously fixed traits
ing found in the fossil record (Eldredge 2003). in the face of moderate gene flow (Nuismer et
Data from several paleontological studies on al. 2000). As a result, the geographic mosaic of
Pleistocene plants (Davis 1983), beetles (Coope coevolution may often create ongoing genetic
1979), foraminifera (Bennett 1990), and mol- dynamics embedded within longer-term sta-
lusks (Valentine and Jablonski 1993) have sis, with populations only rarely breaking
demonstrated little morphological response to through in fundamentally novel directions.
protracted climate change. Instead, geograph- Increasingly detailed paleontological stud-
ic distributions changed. Species tended to ies corroborate the potential importance of
survive, usually with little or no discernible spatial structure in maintaining stasis. More-
morphological change, as long as recognizable over, the great strength of paleontological data
habitats could be tracked. That does not mean is that within-population variation can be
that natural selection is not acting (Davis and compared over time as well as space, allowing
Shaw 2001; Hoekstra et al. 2001), as the data analysis of the importance of the spatial struc-
from population and evolutionary genetics turing of species throughout a species history.
show that populations are constantly under Analysis of two broadly distributed species
selection. Rather, it means that selection often lineages of Devonian brachiopods highlights
acts in ways that favor populations that are the significance of spatial structuring within
evolutionarily conservative at the species level. species to the generation of patterns of stasis
The geographic mosaic of coevolution may and change (Lieberman et al. 1995). Statistical
also contribute to species-wide stasis, even analysis revealed no discernible net change in
though coevolution is one of the evolutionary the morphology of either species over their re-
forces most commonly thought to generate spective five-million-year histories. However,
novelty. Studies over the past decade have in- within any single environment, large morpho-
dicated that selection mosaics, coevolutionary logical shifts did occur larger, in fact, than
hotspots, and gene flow can combine to create the net morphological change across the entire
extensive coevolutionary dynamics (Thomp- environmental distribution of the species over
son 1994, 1997, 1999a). This ongoing coevo- the same time period (Fig. 3). As these chang-
lution creates local novelty and is undoubt- es were in different directions in different en-
142 NILES ELDREDGE ET AL.
FIGURE 3. Schematic diagram showing temporal and environmental (spatial) patterns of morphological change in
two species of Middle Devonian brachiopods, measured as Mahalanobis D2 values from canonical discriminant
analysis of morphometric data. Each of these species occurred in five distinct environments over a period of 5 Myr.
Note the oscillatory nature of morphological change in each species (left and middle panels). The morphological
changes of Mediospirifer audaculus sampled from the five distinct environments (far right panel) are also oscillatory,
but have larger D2 distance excursions than when samples of the species are lumped as a whole (see middle panel).
Moreover, changes within individual environments tend to cancel out, leading to negligible net change for the spe-
cies as a whole.
vironments, they tended to cancel out, result- stantial geographic sampling. Both of these
ing in no net change: stasis resulted at least in data sets contain examples of short-term evo-
part from a species presence in several dis- lutionary change that is repeatedly reversed
tinct environments (Lieberman et al. 1995; Lie- over longer timescales (Gingerich 1983)
berman and Dudgeon 1996). much like the fluctuations in beak morpholo-
These results agree with the expectation gy in Galapagos finches (Grant 1986) and in
that spatial structuring of widespread species floral color of desert plants (Schemske and
will, as a rule, lead to stasis but that local Bierzychudek 2001). Thus, the entire spatio-
populations, under certain conditions, can be temporal history of a species can reveal less
expected to develop more substantial net change than what is documented in tem-
amounts of morphological change in the short poral or geographic subsets of a species line-
term. They further suggest that the patterns of age.
generally fluctuating change documented by
Conclusions
Gingerich (1976) in Eocene mammals and
Sheldon (1987) in Ordovician trilobites reflect Both theoretical and empirical studies of the
the evolutionary histories of geographically past decade suggest that the complex pattern
localized populations of these species. Gin- of selection imposed on geographically struc-
gerich s data involved meticulously collected tured populations by heterogeneous environ-
time series from the Bighorn Basin of Wyo- ments and coevolution can paradoxically
ming, a localized subset of the regions over maintain stasis at the species level over long
which the Hyopsodus and other species have periods of time. By contrast, neither lack of ge-
been documented to have lived. Likewise, netic variation nor genetic and developmental
Sheldon s study of eight trilobite lineages from constraint is probably sufficient in and of itself
the Builth Inlier of Wales did not include sub- to account for species-wide stasis.
DYNAMICS OF EVOLUTIONARY STASIS 143
Further resolution of our understanding of Literature Cited
the dynamics of evolutionary stasis will re-
Barton, N. H., and B. Charlesworth. 1984. Genetic revolutions,
quire novel integration of modeling and em- founder effects, and speciation. Annual Review of Ecology
and Systematics 15:133 164.
pirical analyses. Comparison of rates of grad-
Barton, N. H., and L. Partridge. 2000. Limits to natural selection.
ual change in widespread versus endemic
BioEssays 22:1075 1084.
Beldade, P., K. Koops, and P. M. Brakefield. 2002. Developmental
species will help us better test our conclusion
constraints versus flexibility in morphological evolution. Na-
that geographic range shapes stasis. Such
ture 416:844 847.
analyses of the genetic and geographic struc- Bennett, K. D. 1990. Milankovitch cycles and their effects on spe-
cies in ecological and evolutionary time. Paleobiology 16:11
ture of species when placed within a phylo-
21.
genetic context will help us further test the rel-
Brakefield, P. M., V. French, and B. J. Zwaan. 2003. Development
ative contributions of geographic structure, and the genetics of evolutionary change within insect species.
Annual Review of Ecology, Evolution and Systematics 34:633
underlying genetic variation, and develop-
660.
ment to the ongoing dynamics of stasis.
Brett, C. E., and G. Baird. 1995. Coordinated stasis and evolu-
tionary ecology of Silurian to Middle Devonian faunas in the
Remaining issues include finer resolution of
Appalachian Basin. Pp. 285 315 in R. Anstey and D. H. Erwin,
the issue of conservation versus generation of
eds. Speciation in the fossil record. Columbia University
novelty in short bursts of speciation, and the
Press, New York.
Burdon, J. J., and P. H. Thrall. 1999. Spatial and temporal pat-
possibility that many such bursts of speciation
terns in coevolving plant and pathogen associations. Ameri-
are spatiotemporally correlated among sym-
can Naturalist 153:S15 S33.
patric lineages in regional ecological settings.
   . 2000. Coevolution at multiple spatial scales: Linum mar-
ginale-Melampsora lini from the individual to the species.
Such bursts could reflect   turnovers  (Vrba
Evolutionary Ecology 14:261 281.
1985) or the events between episodes of   co-
Charlesworth, B., R. Lande, and M. Slatkin. 1982. A neo-Dar-
ordinated stasis  (Brett and Baird 1995), re- winian commentary on macroevolution. Evolution 36:474
498.
flecting spatiotemporal scales intermediate to
Coope, G. R. 1979. Late Cenozoic fossil Coleoptera: evolution,
local ecological succession, on the one hand,
biogeography and ecology. Annual Review of Ecology and
and the well-documented evolutionary re- Systematics 10:247 267.
Cooper, V. S., and R. E. Lenski. 2000. The population genetics of
sponses to episodes of global mass extinctions
ecological specialization in evolving E. coli populations. Na-
on the other (Eldredge 2003). And what dy- ture 407:736 739.
Coyne, J. A., N. H. Barton, and M. Turelli. 1997. A critique of
namic processes underlie the emergence of
Sewall Wright s shifting balance theory of evolution. Evolu-
stable species (Miller 2003)? The solution to
tion 51:643 671.
these and related problems will demand fur- Darwin, C. D. 1871. The descent of man, and selection in relation
to sex. John Murray, London.
ther integration of the fields of evolutionary
Davis, M. 1983. Quaternary history of deciduous forests of east-
ecology and evolutionary developmental bi-
ern North America and Europe. Annals of the Missouri Bo-
ology into evolutionary genetic and paleon- tanical Garden 20:550 563.
Davis, M. B., and R. G. Shaw. 2001. Range shifts and adaptive
tological approaches.
responses to Quaternary climate change. Science 292:673 679.
Ehrlich, P. R., and P. H. Raven. 1964. Butterflies and plants: a
study in coevolution. Evolution 18:586 608.
Acknowledgments
Eldredge, N. 1971. The allopatric model and phylogeny in Pa-
leozoic invertebrates. Evolution 25:156 167.
This work was conducted as part of the Eco-
   . 1989. Macroevolutionary dynamics: species, niches and
logical Processes and Evolutionary Rates
adaptive peaks. McGraw-Hill, New York.
   . 2003. The sloshing bucket: how the physical realm con-
Working Group supported by National Center
trols evolution. Pp. 3 32 in J. Crutchfield and P. Schuster, eds.
for Ecological Analysis and Synthesis at Uni-
Evolutionary dynamics: exploring the interplay of selection,
versity of California, Santa Barbara (NSF accident, neutrality, and function (SFI Studies in the Sciences
of Complexity Series). Oxford University Press, New York.
DEB-94-21535); and by National Science Foun-
Eldredge, N., and S. J. Gould. 1972. Punctuated equilibrium: an
dation support to J.N.T., D.J., R.E.L., B.S.L.,
alternative to phyletic gradualism. Pp. 82 115 in T. J. M.
M.A.M., S.G.; Human Frontiers Science Pro- Schopf, ed. Models in paleobiology. Freeman, Cooper, San
Francisco.
gram support to P.M.B.; and National Insti-
Falconer, D. S., and T. Mackay. 1996. Introduction to quantitative
tutes of Health support to S.G. We thank D. J.
genetics, 4th ed. Longman, London.
Futuyma, D. J. 1987. On the role of species in anagenesis. Amer-
Futuyma for discussion, and C. Thomas, J.
ican Naturalist 130:465 473.
Valentine, and anonymous reviewers for com-
Gandon, S., Y. Capowiez, Y. Dubois, Y. Michalakis, and I. Oli-
ments on the manuscript. vieri. 1996. Local adaptation and gene-for-gene coevolution in
144 NILES ELDREDGE ET AL.
a metapopulation model. Proceedings of the Royal Society of stasis and change in two species lineages from the Middle De-
London B 263:1003 1009. vonian of New York State. Paleobiology 21:15 27.
Garcia-Ramos, G., and M. Kirkpatrick. 1997. Genetic models of Lynch, M., and A. Force. 2000. The probability of duplicate gene
adaptation and gene flow in peripheral populations. Evolu- preservation by subfunctionalization. Genetics 154:459 473.
tion 51:21 28. Mani, G. S., and B. C. C. Clarke. 1990. Mutational order: a major
Gavrilets, S. 1996. On phase three of the shifting-balance theory. stochastic process in evolution. Proceedings of the Royal So-
Evolution 50:1034 1041. ciety of London B 240:29 37.
   . 1997. Evolution and speciation on holey adaptive land- Maynard Smith, J. 1983. The genetics of stasis and punctuation.
scapes. Trends in Ecology and Evolution 13:307 312. Annual Review of Genetics 17:11 25.
   . 1999. A dynamical theory of speciation on holey adap- Mayr, E. 1954. Change of genetic environment and evolution.
tive landscapes. American Naturalist 154:1 22. Pp. 157 180 in J. Huxley, A.C. Hardy, and E. B. Ford, eds. Evo-
Geary, D. H. 1995. The importance of gradual change in species- lution as a process. Allen and Unwin, London.
level transitions. Pp. 67 86 in D. H. Erwin and R. L. Anstey, McPeek, M. A. 1997. Measuring phenotypic selection on an ad-
eds. New approaches to speciation in the fossil record. Co- aptation: lamellae of damselflies experiencing dragonfly pre-
lumbia University Press, New York. dation. Evolution 51:459 466.
Gingerich, P. D. 1976. Paleontology and phylogeny: patterns of    . 1998. The consequences of changing the top predator in
evolution at the species level in Early Tertiary mammals. a food web: a comparative experimental approach. Ecological
American Journal of Science 276:1 28. Monographs 68:1 23.
   . 1983. Rates of evolution: effects of time and temporal    . 2000. Predisposed to adapt: clade-level differences in
scaling. Science 222:159 161. characters affecting swimming performance in damselflies.
Gomulkiewicz, R., and R. D. Holt. 1995. When does evolution by Evolution 54:2072 2080.
natural selection prevent extinction? Evolution 49:201 207. McPeek, M. A., and J. M. Brown. 2000. Building a regional spe-
Gomulkiewicz, R., J. N. Thompson, R. D. Holt, S. L. Nuismer, cies pool: diversification of the Enallagma damselflies of east-
and M. E. Hochberg. 2000. Hot spots, cold spots, and the geo- ern North American waters. Ecology 81:904 920.
graphic mosaic theory of coevolution. American Naturalist Miller, W., III. 2003. A place for phyletic evolution within the
156:156 174. theory of punctuated equilibria: Eldredge pathways. Neues
Gould, S. J., and N. Eldredge. 1977. Punctuated equilibrium: the Jahrbuch für Geologie und Paläontologie Monatshefte 2003:
tempo and mode of evolution reconsidered. Paleobiology 3: 463 476.
115 151. Mongold, J. A., A. F. Bennett, and R. E. Lenski. 2001. Evolution-
Grant, P. R. 1986. Ecology and evolution of Darwin s finches. ary adaptation to temperature. VII. Extension of the upper
Princeton University Press, Princeton, N.J. thermal limit of Escherichia coli. Evolution 53:386 394.
Hanski, I., and M. E. Gilpin. 1997. Metapopulation biology: ecol- Moore, F. B.-G., D. E. Rozen, and R. E. Lenski. 2000. Pervasive
ogy, genetics, and evolution. Academic Press, San Diego. compensatory adaptation in Escherichia coli. Proceedings of
Hoekstra, H. E., J. M. Hoekstra, D. Berrigan, S. N. Vignieri, A. the Royal Society of London B 267:515 522.
Hoang, C. E. Hill, P. Beerli, et al. 2001. Strength and tempo of Nuismer, S. L., J. N. Thompson, and R. Gomulkiewicz. 2000. Co-
directional selection in the wild. Proceedings of the National evolutionary clines across selection mosaics. Evolution 54:
Academy of Sciences USA 98:9157 9160. 1102 1115.
Huey, R. B., G. W. Gilchrist, M. L. Carlson, D. Berrigan, and L. Ohta, T. 1972. Population size and rate of evolution. Journal of
Serra. 2000. Rapid evolution of a geographic cline in size in Molecular Evolution 1:305 314.
an introduced fly. Science 287:308 309. Pellmyr, O., J. Leebens-Mack, and J. N. Thompson. 1998. Her-
Jablonski, D. 2000. Micro- and macroevolution scale and hier- bivores and molecular clocks as tools in plant biogeography.
archy in evolutionary biology and paleobiology. In D. H. Er- Biological Journal of the Linnean Society 63:367 378.
win and S. L. Wing, eds. Deep time: Paleobiology s perspective. Reznick, D. N., F. H. Shaw, F. H. Rodd, and R. G. Shaw. 1997.
Paleobiology 26(Suppl. to No. 4):15 52. Evaluation of the rate of evolution in natural populations of
Jackson, J. B. C., and A. H. Cheetham. 1999. Tempo and mode guppies (Poecilia reticulata). Science 275:1934 1936.
of speciation in the sea. Trends in Ecology and Evolution 14: Rieseberg, L. H. 1997. Hybrid origins of plant species. Annual
72 77. Review of Ecology and Systematics 28:359 389.
Kawata, M. 2002. Invasion of vacant niches and subsequent Rutherford, S. L., and S. Lindquist. 1998. Hsp90 as a capacitor
sympatric speciation. Proceedings of the Royal Society of for morphological evolution. Nature 396:336 342.
London B 269:55 63. Sandstrom, J. P., J. A. Russell, J. P. White, and N. A. Moran. 2001.
Kirkpatrick, M., and N. H. Barton. 1997. Evolution of a species Independent origins and horizontal transfer of bacterial sym-
range. American Naturalist 150:1 23. bionts of aphids. Molecular Ecology 10:217 228.
Lambeck, K., and J. Chappell. 2001. Sea level change through Schemske, D. W., and P. Bierzychudek. 2001. Evolution of flower
the last glacial cycle. Science 292:679 686. color in the desert annual Linanthus parryae: Wright revisited.
Lande, R. 1985. The fixation of chromosomal rearrangements in Evolution 55:1269 1282.
a subdivided population with local extinction and coloniza- Sheldon, P. R. 1987. Parallel gradualistic evolution of Ordovician
tion. Heredity 54:323 332. trilobites. Nature 330:561 563.
Lenski, R. E., and M. Travisano. 1994. Dynamics of adaptation Sniegowski, P. D., P. J. Gerrish, and R. E. Lenski. 1997. Evolution
and diversification: a 10,000- generation experiment with bac- of high mutation rates in experimental populations of Esche-
terial populations. Proceedings of the National Academy of richia coli. Nature 387:703 705.
Sciences USA 91:6808 6814. Soltis, D. E., and P. S. Soltis. 1999. Polyploidy: recurrent for-
Levinton, J. S. 1983. Stasis in progress: the empirical basis of mation and genome evolution. Trends in Ecology and Evo-
macroevolution. Annual Review of Ecology of Systematics 14: lution 14:348 352.
103 137. Stanley, S. M., and X. Yang. 1987. Approximate evolutionary sta-
Lieberman, B. S., and S. Dudgeon. 1996. An evaluation of sta- sis for bivalve morphology over millions of years: a multivar-
bilizing selection as a mechanism for stasis. Palaeogeography, iate, multilineage study. Paleobiology 13:113 1119.
Palaeoclimatology and Palaeoecology 127:229 238. Tachida, H., and M. Ilizuka. 1991. Fixation probability in spa-
Lieberman, B. S., C. E. Brett, and N. Eldredge. 1995. A study of tially changing environments. Genetical Research 58:243 251.
DYNAMICS OF EVOLUTIONARY STASIS 145
Thomas, C. D., E. J. Bodsworth, R. J. Wilson, A. D. Simmons, Z. Van Valen, L. M. 1982. Integration of species: stasis and bioge-
G. Davies, M. Musche, and L. Conradt. 2001. Ecological and ography. Evolutionary Theory 6:99 112.
evolutionary processes at expanding range margins. Nature Vrba, E. S. 1985. Environment and evolution: alternative causes
411:577 581.
of the temporal distribution of evolutionary events. South Af-
Thompson, J. N. 1994. The coevolutionary process. University of
rican Journal of Science 81:229 236.
Chicago Press, Chicago.
Wade, M. J., and C. J. Goodnight. 1998. The theories of Fisher
   . 1997. Evaluating the dynamics of coevolution among
and Wright in the context of metapopulations: when nature
geographically structured populations. Ecology 78:1619
does many small experiments. Evolution 52:1537 1553.
1623.
Wake, D. B., G. Roth, and M. H. Wake. 1983. On the problem of
   . 1998. Rapid evolution as an ecological process. Trends
stasis in organismal evolution. Journal of Theoretical Biology
in Ecology and Evolution 13:329 332.
101:211 224.
   . 1999a. The evolution of species interactions. Science 284:
Weber, K. E. 1996. Large genetic change at small fitness cost in
2116 2118.
large populations of Drosophila melanogaster selection for wind
   . 1999b. Coevolution and escalation: are ongoing coevo-
tunnel flight: rethinking fitness surfaces. Genetics 144:205
lutionary meanderings important? American Naturalist 153:
213.
S92 S93.
Webster, A. J., R. J. H. Payne, and M. Pagel. 2003. Molecular phy-
Thrall, P. H., and J. J. Burdon. 1997. Host-pathogen dynamics in
logenies link rates of evolution and speciation. Science 301:
a metapopulation context: the ecological and evolutionary
478.
consequences of being spatial. Journal of Ecology 85:743 753.
Williamson, P. G. 1987. Selection or constraint? A proposal on
Travisano, M., J. A. Mongold, A. F. Bennet, and R. E. Lenski.
the mechanism for stasis. Pp. 129 142 in K. S. W. Campbell
1995. Experimental tests of the roles of adaptation, chance,
and M. F. Day, eds. Rates of evolution. Allen and Unwin, Lon-
and history in evolution. Science 267:87 90.
Valentine, J. W., and D. Jablonski. 1993. Fossil communities: com- don.
Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups. 2001.
positional variation at many time scales. Pp. 341 349 in R. E.
Ricklefs and D. Schluter, eds. Species diversity in ecological Trend, rhythms, and aberrations in global climates 65 Ma to
communities. University of Chicago Press, Chicago. present. Science 292:686 693.