AMER. ZOOL., 32:123-134 (1992)
Rates of Evolution in Developmental Processes1
GREGORY A. WRAY
Friday Harbor Laboratories and Department of Zoology, University of Washington,
620 University Road, Friday Harbor, Washington 98250
SYNOPSIS. The tempo and mode of morphological evolution are influenced by several
factors, among which evolutionary transformations in developmental processes are likely
to be important. Comparing the embryos of extant species in an explicit phylogenetic framework allows the estimation of minimum average rates of evolution in quantitative developmental parameters. It also allows delineation of the maximum time that complex qualitative transformations in developmental mechanism take to evolve. This paper analyzes
rates of quantitative and qualitative developmental evolution using examples drawn primarily from echinoderms. The results demonstrate that rates of developmental evolution
can be comparable to rates of morphological evolution. There is no indication that rates of
evolution in development are lower for earlier stages, contrary to the prediction of "tree"
models of epigenetic interactions. In particular, rates of evolution in oogenesis can exceed
rates of evolution in adult body size. Rates of developmental evolution can vary by up to
two orders of magnitude within a clade. Whether such large scale variation in evolutionary
rates of developmental processes is a general phenomenon can only be answered by further
study.
in developmental processes. Such an
endeavor
would be interesting because
From the outset, rates of change have
formed a central theme in studies of mac- development plays a central role in evoluroevolution (Simpson, 1944, 1953; Hal- tion, mechanistically linking genomic and
dane, 1949; Stanley, 1979). Evolutionary morphological change. It therefore seems
rates remain an important component in likely that rates of change in developmental
contemporary macroevolution, as attested process will provide insights into evoluby the interest generated by the punctuated tionary rate phenomena. For example,
equilibrium debate and the growing atten- development can dissociate rates of genotion on rates of genomic change. Evolu- mic and phenotypic change by amplifying
tionary rate studies have focused on differ- or buffering the effects of mutations. Of parences in rates of evolution between clades ticular interest is the possibility that rates
and on uneven rates of evolution within of developmental evolution may constrain
clades (Simpson, 1944; Stanley, 1979,1985). or boost rates of change in morphology.
Differences in rates of evolution may influThis paper explores tempo and mode in
ence clade diversity, shape, and persistence the evolution of development. I begin by
by affecting speciation rates and the origin providing methods for reconstructing and
of evolutionary innovations (Simpson, 1944; dating evolutionary transformations in
Eldredge and Gould, 1972; Van Valen, 1974; development. Following this approach, it is
Stanley, 1979).
possible to estimate minimum average rates
Historically, most evolutionary rate stud- of evolution in quantitative developmental
ies have focused on change in adult mor- parameters, as well as the duration that
phology, although increasing attention is complex qualitative transformations in
now being directed towards rates of molec- developmental processes take to evolve.
ular evolution. To date, however, no attempt Using these methods I then demonstrate that
has been made to assess rates of evolution development processes can evolve at rates
comparable to adult morphology, and that
rates of evolution in development can vary
1
From the Symposium on Development and Macby over two orders of magnitude within a
roevolution sponsored by the Division of the History clade. Finally, I consider the relevance of
and Philosophy of Biology of the American Society of
variable rates of change in developmental
Zoologists and presented at the Annual Meeting of the
processes to macroevolution, including the
American Society of Zoologists, 27-30 December 1990,
at San Antonio, Texas.
decoupling of genomic and morphological
INTRODUCTION
123
124
GREGORY A. WRAY
rates of change, the possibility of stage-specific rates of developmental evolution, and
the extent to which developmental processes may limit evolutionary changes in
body plans.
RECONSTRUCTING THE
EMBRYOS OF ANCESTORS
Although the embryos of extinct ancestors cannot be studied directly, there is much
that can be learned about them from their
living descendants. An analysis of character
evolution (Maddison et ai, 1984; Swofford
and Maddison, 1987; Donoghue, 1989)
provides a means of reconstructing ancestral ontogenies: differences in the developmental processes of extant species are
mapped onto an independently derived
phylogeny, and ancestral states at each node
are inferred using biologically reasonable
parsimony assumptions. As with variation
in any other trait, outgroups can be used to
polarize character state transformations
(Farris, 1982; Swofford and Maddison,
1987). When combined with estimated
divergence times, this type of analysis can
be used to date specific evolutionary transformations in development (Wray and Raff,
1991a).
As an example, consider the evolution of,
developmental processes within the echinoderm class Echinoidea (sea urchins, sand
dollars, and their kin). Features of early
echinoid development such as radial, holoblastic cleavage, and regulative early cell
divisions are symplesiomorphies for the
phylum Echinodermata; they are present in
extant echinoids and were almost certainly
present in the earliest echinoids as well (Fig.
1, traits 1-7). There are several derived features of development shared by phylogenetically diverse extant echinoids that are
not present in other echinoderm classes
(Fig. 1, traits 8-10). These features, such as
unequal fourth cleavage divisions and the
characteristic echinopluteus larva, were
therefore likely present in the echinoid crown
group ancestor. Evolutionary transformations can also be traced within the crown
group itself: cidaroids are distinguished by
late ingression of skeletogenic cells and loss
of the apical tuft of cilia (Schroeder, 1981;
Wray and McClay, 1988) (Fig. 1, traits 11
and 12), while euechinoids have maternal
transcription of a-histone (Raff et ai, 1984)
and a characteristic distribution of skeletogenic cells during gastrulation (Okazaki,
1975) (Fig. 1, traits 13 and 14). Polarizing
these features to outgroups allows a reconstruction of development in the crown group
ancestors of the cidaroids and euechinoids.
Using dates for divergence times, it is possible to estimate the minimum period a particular developmental process has persisted
within a given lineage (Fig. 1).
MEASURING RATES OF
EVOLUTION IN DEVELOPMENT
Evolutionary changes in development
take many forms and affect a wide variety
of processes. As explained below, however,
it is only necessary to distinguish between
quantitative and qualitative transformations when estimating rates of evolution in
developmental processes. Because modifications can evolve in developmental processes without significantly affecting adult
morphology (Elinson, 1987; Strathmann,
1988a; Raff et ai, 1991), rates of morphological evolution in adults can underestimate rates of evolution in development.
Estimating evolutionary rates in
quantitative parameters
Much about development is inherently
quantitative, including parameters of size,
rate, and time. Quantitative change, particularly in the form of heterochrony, is a common component in the evolution of development (McNamara, 1988; Raff and Wray,
1989). For the purposes of comparison
between taxa, it is preferable to calculate
proportional, rather than absolute, rates of
evolution in quantitative features (Haldane,
1949; Simpson, 1953; Van Valen, 1974).
The minimum average proportional rate of
evolution, r, in a quantitative parameter, P,
is:
_ _ 1 dV _ ln(P de J - ln(Panc)
r
"PdT
i
(1)
where Panc and Pdesc are the values of the
parameter in an ancestor and a descendant,
and where t is the elapsed time (Haldane,
1949).
125
RATES OF DEVELOPMENTAL EVOLUTION
r =
FIG. 1. Dating evolutionary transitions in echinoid
development. Differences in the development of echinoids are mapped onto a cladogram simplified from
Smith (1984). Traits 1-7 are symplesiomorphies for
the phylum and therefore at least 590 Ma old (Smith,
1988a): (1) radial cleavage, (2) holoblastic cleavage, (3)
first two cleavage divisions regulative, (4) gastrulation
by invagination followed by cell rearrangement, (5)
adult hydrocoel derived from the left larval mesocoel,
(6) monociliated band surrounding the larval mouth,
and (7) planktotrophic development. Traits 8-10 arose
between the divergence of echinoids from their sister
group (approximately 510 Ma ago; Smith, 1988a) and
the radiation of the crown group (approximately 250
Ma ago; Kier, 1984; Smith, 1984): (8) oocyte arrest
after meiosis II, (9) unequal vegetal fourth and fifth
cleavage divisions, and (10) an echinoid-specific larval
skeleton. Trait 11 is reconstructed as derived in cidaroids (absence of an apical tuft of cilia in the larva),
while trait 12 appears to represent a reversal (ingression
of skeletogenic cells during gastrulation). Two traits are
hypothesized as derived in euechinoids: (13) maternal
synthesis of a-histone and (14) subequatorial ring of
skeletogenic cells during gastrulation. For sources of
character distributions, see Kume and Dan, 1968;
Emlet, 1988; Strathmann, 1988a; and Wray and Raff,
19916.
For developmental parameters, however,
we usually do not have a direct measure of
P in the ancestor, and this value must be
inferred from extant species. The most conservative assumption is that the value of P
in the ancestor was such that equal and
opposite rates of evolution produced the
values in two extant descendants. If we were
concerned here with absolute rates of evolution, this value would be the arithmetic
mean of that in the two descendants (Fig.
2A). The comparable value for proportional
rates of change is the geometric mean, which
is smaller than the arithmetic mean (Fig.
2B). The value of r can thus be estimated
from values of P in two extant species as:
P2)/2]
(2)
where td is the estimated divergence time
between the two extant species and P, > P2
(see Appendix for derivation).
It is important to note that r is a conservative estimate for rates of evolution. It
underestimates true evolutionary rates
where the value of the parameter in the
ancestor was not the proportional mean of
that in the extant species (Fig. 1C), and where
rates of change between ancestor and
descendant are nonlinear (Fig. ID). These
conditions, especially the latter, are probably common, and the true rate of evolution
will therefore often exceed r. Estimates of r
will be most meaningful where the two
extant species used to derive the values of
P are closely related, reducing the possibility
of changes, or even reversals, in evolutionary rate.
As an example, consider evolutionary
changes in echinoderm oogenesis. Egg volume is a quantitative feature of development with ecological and evolutionary consequences (Strathmann, 1985; Jablonski,
1986; Emlet et al, 1987). Egg volumes in
the extant sea stars Oreaster reticulatus and
O. occidentalis are 4.00 x 106/uni3and 1.59
x 106 Mm3 respectively (Lessios, 1990).
These species diverged 3.1-3.5 Ma ago during a major vicariance event, the elevation
of the Panamanian Isthmus (Keigwin, 1978,
1982). The latest common ancestor is
reconstructed as having an egg with a volume of 2.52 x 106 jum3. Conservatively
assuming the longest divergence time, r in
egg volume within Oreaster is at least 0.13
Ma"1. This is a rate of approximately 49
millidarwins (1 millidarwin represents an
increase or decrease in size by a factor of e
per Ma; Haldane, 1949). This value falls
within the range of "normal" rates of evolution in adult body size, a point considered
in more detail later.
By making several pairwise comparisons
between extant species, it is possible to estimate the range of rates of evolution in a
quantitative developmental parameter
within a single clade. Table 1 lists several
estimates for r in echinoderm oogenesis,
based on six intrageneric comparisons. Note
that low and high estimates for r can differ
126
GREGORY A. WRAY
B
A
anc
anc
D
'anc
anc
FIG. 2. Estimating minimum proportional rates of evolution. The value of a quantitative parameter, P, is
plotted against evolutionary time. The arithmetic mean of P in two extant species will not yield equal rates of
change when calculating proportional rates of evolution (A), but the geometric mean will (B) (see Appendix).
Using the geometric mean as the reconstructed ancestral value will therefore produce minimum estimates for
proportional rates of evolution. In most real situations, rate calculations based on the geometric mean will
underestimate actual rates of evolution; this will occur where Panc was not the geometric mean (C) or where
evolutionary change was nonlinear (D).
substantially between even closely related
taxa: within the genus Strongylocentrotus or
the subfamily Echinometrinae (Tripneustes
vs. Heliocidaris). Thus minimum estimates
for net rates of change in a developmental
process such as oogenesis can be substantially lower than even conservative estimates for maximal rates of change. We will
return to this point later.
Estimating the duration of qualitative
transformations in development
Many evolutionary changes in development are inherently qualitative, and involve
discontinuities in developmental mechanism. The appearance of a novel developmental process or the loss of an existing one
is a qualitative transformation, as are evolutionary substitutions in developmental
mechanism. The simplest kind of qualitative transformation is one resulting from a
single mutation. In such cases, no intermediate states exist and individuals display
either the ancestral or the derived condition. Evolutionary transformations in
development resulting from a single mutation are effectively instantaneous, and rates
of change are meaningless.
There are, however, many qualitative
transformations in development where a
single mutation scenario is highly unlikely.
In such cases, a functionally integrated suite
of changes transforms an evolutionarily
conservative ancestral state into a qualita-
127
RATES OF DEVELOPMENTAL EVOLUTION
TABLE 1. Rates of evolution in egg volume among
echinoderms.
Subject
V tMa]
Proportional
rate of change
[millidarwins]'
[Ma-]"
Oreastei"
Strongylocentrotus'
Heliocidaris*
Strongylocentrotus'
Tripneustes*
Diadema}
3.5'
2O
8'
2O
3.5'
3.5'
0.132
0.045
0.293
0.005
0.005
0.007
49
17
108
2
2
3
• Divergence times in millions of years; rate calculations are based on maximum estimates for divergence times.
b
Proportional rates of change per million years calculated according to equation (2).
'Proportional rate of evolution in millidarwins (1
millidarwin = a change in size by a factor of e per
million years; Haldane, 1949).
d
O. reticulatus 4.00 x 106 jim3 and O. occidentalis
1.59 x 106/mi3(Lessios, 1990).
• Based on uplift of Panamanian isthmus, 3.1-3.5 Ma
ago (Keigwin, 1982).
f
5. purpuratus 0.30 x 106 jim3 and S. droebachiensis
1.84 x 106 (im' (Strathmann, 1987).
' Based on fossil record, 3.5-20 Ma ago (Smith, 19886).
h
H. erythrogramma 41.63 x 106 ^m3 and H. tuberculata 0.38 x JO6 Mm3 (Raff, 1987).
' Based on mitochondrial DNA restriction site polymorphisms, 6-8 Ma ago (McMillan et al., 1992).
' S. droebachiensis 1.84 x 106 (im! and S. pallidus
1.50 x 106 Mm3 (Strathmann, 1987).
k
T. ventricosus 0.27 x 106Mm3and T. depressus 0.26
x 106 >im3 (Lessios, 1990).
1
D. antillarum 1.57 x 105 nm3 and D. mexicanum
1.69 x 105/xm3 (Lessios, 1990).
tively derived condition through a series of
intermediate states. By reconstructing the
development of ancestors with known
divergence times, it is possible to obtain a
conservative estimate of the maximum
duration a complex qualitative change took
to evolve. Similarly, one can estimate the
minimum duration the ancestral state has
persisted from the estimated divergence
times of extant species sharing this character state. In this way it is possible to distinguish between gradual and rapid qualitative transformations in development.
Transitions in developmental mode provide dramatic examples of complex, functionally integrated, qualitative changes in
ontogeny. The transformation from feeding
to non-feeding larval development is a
recurrent life history alteration in many animal phyla (Jagersten, 1972; Strathmann,
1985), and one that often involves substan-
> 590 Ma
FIG. 3. Dating the origin of lecithotrophic development in Heliocidaris. Although planktotrophic development is characteristic of echinometrid sea urchins,
Heliocidaris erythrogramma has direct development
(Raff, 1987). Lecithotrophic development is reconstructed as the derived condition by successive outgroup comparison. H. erythrogramma diverged from
its congener H. tuberculata about 6-10 Ma ago
(McMillan et al., 1992), providing a lower bound on
the origin of lecithotrophic development. Two populations of//, erythrogramma, both with lecithotrophic
development, diverged 2 Ma ago, providing an upper
bound to this transformation. A conservative estimate
for the time lecithotrophic development took to evolve
is thus 4-8 Ma. Planktotrophic development is a symplesiomorphic feature of echinoderms as a phylum,
and is present in outgroups such as enteropneust hemichordates (Jagersten, 1972; Strathmann, 1978, 1988a,
b). A conservative estimate for the persistence of the
ancestral state is thus 590 Ma, the age of the oldest
echinoderm fossils (Smith, 1988a). The within-phylum ancestors of Heliocidaris almost certainly were all
indirect developers, as there is no evidence of a secondary derivation for indirect development within the
phylum (Strathmann, 1988a).
tive modifications in developmental mechanism (Elinson, 1987; Jeffery and Swalla,
1990; Wray and Raff, 1991 a). The sea urchin
genus Heliocidaris provides a particularly
well-studied example of a highly complex,
qualitative transformation in development,
and illustrates the diversity of developmental processes that can be modified in concert
(Wray and Raff, 1990, 1991a).
H. tuberculata exhibits planktotrophic
(indirect) development, while H. erythrogramma has lecithotrophic (direct) development (Raff, 1987); other urchins from the
same family and from related families are
planktotrophic developers, so lecithotrophic development is reconstructed as the
derived condition (Fig. 3). Divergence times
within this genus have been calculated from
128
GREGORY A. WRAY
mitochondrial DNA restriction site poly- course, the very notion of a "normal" rate
morphisms (McMillan et al, 1992). Two of evolution may be misleading, since evopopulations of H. erythrogramma with lutionary change may be strongly punctuadirect development separated approxi- tional (Hallam, 1978; Stanley and Yang,
mately 2 Ma ago, while H. tuberculata and 1987).
H. erythrogramma diverged approximately
The rates of evolution in echinoderm
6-10 Ma ago. A conservative upper esti- oogenesis presented in Table 1 cover a broad
mate of the period required for the evolu- range of values. Significantly, the range of
tion of direct development in Heliocidaris evolutionary rates for echinoderm egg sizes
is thus 8 Ma (Fig. 3).
overlaps broadly with published rates of
The transition from planktotrophic to evolution for adult morphology in marine
lecithotrophic development appears to be invertebrates (Van Valen, 1974; Hallam,
unidirectional in many metazoan phyla, 1975; Gingerich, 1983; Stanley and Yang,
including echinoderms (Jagersten, 1972; 1987). The evolutionary rate within HelioStrathmann, 1988a). Further, echinoderms cidaris is particularly fast, while, at the other
as a phylum almost certainly arose from an end of the spectrum, there are several rates
ancestor with planktotrophic development ofjust a few millidarwins. These slower rates
(Jagersten, 1972; Raff, 1987; Strathmann, are near stasis (Stanley and Yang, 1987).
1988a). It is thus highly likely that all the Thus rates of evolution in echinoderm egg
within-phylum ancestors of Heliocidaris had size span the range from near stasis to rapid
planktotrophic development. Definitive change by the standards of adult morphoechinoderm fossils are present in lower logical evolution. Although it cannot be
Cambrian deposits 590 Ma old (Smith, quantified in the same manner, the devel1988a), providing a conservative estimate opmental processes underlying larval form
of the persistence of the ancestral condition. show the same trend at a qualitative level:
From this example, it is clear that complex hundreds of millions of years of little net
qualitative transformations in developmen- change followed by relatively rapid change
tal processes can evolve very quickly rela- (Fig. 3).
tive to the persistence of the ancestral conMode
dition.
By contrasting low and high average rates
of
change, it is possible to assess whether
VARIATION IN RATES OF
evolutionary
rates in developmental proDEVELOPMENTAL EVOLUTION
cesses are relatively constant (a gradualistic
Tempo
mode) or highly variable (an episodic or
Rates are relative measures, and we would punctuated mode). Although different
like to be able to compare rates of evolution approaches are required when estimating
in developmental processes with rates of rates of evolution in quantitative and qualmorphological evolution. Unfortunately, itative transitions, in both cases it is posthere is little agreement as to what consti- sible to obtain conservative indices of the
tutes an average or typical rate of morpho- disparity between maximal and minimal
logical evolution. Although some authors rates of change within a clade. The ratio
have suggested that "normal" rates of evo- between these estimates is a measure of
lution for marine invertebrates are on the evenness in evolutionary rates.
order of 75 millidarwins or more (Van
Where there is little variation in evoluValen, 1974; Hallam, 1975; Gingerich, tionary rates, the ratio between maximal and
1983), it seems likely that 10 millidarwins minimal rates of change will be near 1.
is a more reasonable estimate (Stanley, 1985; Ratios near 1 thus indicate either stasis or
Stanley and Yang, 1987). For comparison, sustained, linear, unidirectional change.
"normal" evolutionary rates for vertebrate Ratios greater than 1 indicate uneven rates
adult morphology appear to fall roughly in of evolution: that a feature has not evolved
the 20-40 millidarwin range (Haldane, 1949; in a gradual mode and that it can change
Van Valen, 1974; MacFadden, 1985). Of faster than it actually has changed. Whether
RATES OF DEVELOPMENTAL EVOLUTION
A
Cen.
Ma
Mesozoi
u
590:8
B
350:2
C
590:1
o100 -
200 -
300 -
leozo:
u
400 -
(2
500 -
600 -
FIG. 4. Persistence and change in development mode.
An index of evolutionary rate fluctuations is the discrepancy between persistence and change in complex
qualitative aspects of development. Where the ancestral state (open bars) has persisted much longer than
the period during which the derived state arose (closed
bars), rates of change have been uneven. A. Lecithotrophic development evolved in less than 8 Ma in the
echinoid genus Heliocidaris, but the ancestral, planktotrophic mode precedes the origin of the phylum over
590 Ma ago (see Fig. 3). B. Direct development in
salamanders can evolve in less than 1-2 Ma (Daugherty
et al., 1983; Larson, 1984). The ancestral mode of
development (aquatic, feeding larvae) has likely persisted in the lineage leading to extant anurans since the
colonization of land by vertebrates some 350 Ma ago.
C. Lecithotrophic development can evolve in less than
1 Ma in molluscs (Hansen, 1982). The ancestral condition (planktotrophic development) dates back to at
least the lower Cambrian. The discrepancy between
lengthy persistence of ancestral developmental processes and the rapid transitions to derived states indicates sharply uneven rates of evolution in development
for these three cases.
the ratio between maximal and minimal
rates of developmental evolution is biologically significant will depend largely upon
the accuracy of estimates for divergence
times. Where the difference between diver-
129
gence times is larger than the uncertainties
associated with their estimation, it will be
possible to make meaningful estimates of
unevenness in evolutionary rates.
It is important to note that this is an
inherently conservative method for assessing mode in evolution. There are two reasons that actual differences in evolutionary
rates of development might be higher. First,
the data base for developmental rate comparisons is limited, and wider sampling
might reveal examples of longer persistence
times as well as instances of more rapid
change. And second, maximal rate estimates are themselves conservative. For
quantitative changes, actual rates are underestimated where changes are nonlinear and
where the ancestral condition is not the geometric mean of that in the extant species
(Fig. 2C, D). For qualitative changes, the
duration during which a complex developmental transformation arose may be
much shorter than the period bracketed by
divergences of extant representatives.
The examples presented earlier in this
paper demonstrate that developmental processes can evolve in a strikingly non-gradual
manner. For the oogenesis examples (Table
1), minimal and maximal rates of change
within a genus can differ by almost an order
of magnitude (Strongylocentrotus) and
within a subfamily can differ by a factor of
54 (Heliocidaris vs. Tripneustes). The developmental mode example also indicates
uneven rates of developmental evolution: a
ratio of 70 between the persistence of the
ancestral condition and the period during
which the derived condition evolved. Evolutionary transitions in developmental mode
in other phyla exemplify ratios of minimal
to maximal evolutionary rates of 175 and
590 (Fig. 4).
Differences in rates of one to two orders
of magnitude indicate highly uneven rates
of developmental evolution. The foregoing
examples (Table 1, Fig. 4) indicate that some
developmental mechanisms are subject to
episodic changes that greatly exceed normal
net rates of change. Significantly, transformations in developmental mode demonstrate that a diverse array of developmental
mechanisms can evolve at markedly uneven
rates (Wray and Raff, 1991a).
130
GREGORY A. WRAY
RELEVANCE TO MACROEVOLUTION
The relationships between rates of evolution in the genome, in developmental process, and in morphology are complex. In
some cases, development may contribute to
the decoupling of evolutionary changes in
genes and morphology. In addition, many
developmental processes can evolve without affecting adult morphology (Elinson,
1987; Strathmann, 1988a; Wray and
McClay, 1989; Raff et al, 1991). This
empirical evidence violates the predictions
of some hypotheses about the way development evolves.
Developmental buffering and
amplification
Rates of evolution in the genome and in
morphology can be strikingly disjoint
(Cherry et al, 1978; Lessios, 1981; Baverstock and Adams, 1987). There is no reason
to doubt that, fundamentally, change in the
genome underlies change in morphology, but
it is clear that rates of evolution in the two
can be strongly non-linear. Developmental
processes can contribute significantly to this
non-linear relationship by masking or
amplifying the phenotypic consequences of
a mutation.
In the first instance, development acts as
a buffer, screening changes in the genome
from having an effect on morphology (Foote
and Cowie, 1988). Buffering can occur when
developmental processes are functionally
dissociated, such that modifications in one
process do not affect others (Needham, 1933;
Raff and Wray, 1989). The presence of alternative developmental pathways and homeostatic mechanisms can also buffer genomic change (Katz, 1983). The buffering of
phenotype from genetic change over evolutionary time scales has been variously
termed burden (Reidel, 1978), developmental constraint (Maynard Smith et al,
1985), canalization (Waddington, 1957), and
the epigenetic ratchet (Levinton, 1988).
On the other hand, interaction between
developmental processes can amplify even
the smallest mutation (a single base substitution) into a profound and multifaceted
change in morphology. This is because there
is rarely a one-to-one relationship between
genes and morphology: most genes are
pleiotropic and most aspects of phenotype
are polygenic (Simpson, 1944; Wright, 1980;
Nijhout, 1990).
Estimating the relative influence of the
buffering and amplifying effects of development on genotypic change is problematic.
It seems likely, however, that both effects
can influence rates of morphological evolution. Clades that exhibit "regulative"
embryogenesis or a capacity for regeneration may have a greater ability to buffer
genetic change than do clades with less
developmental flexibility. As a consequence, greater intraspecific and interspecific variation in the genetic basis for development may be tolerated in developmentally
flexible clades.
Stage-specific rates of
evolution in development
"Tree" models of epigenetic interactions
predict that a mutation altering early development will have a larger phenotypic consequence than one affecting later development (Arthur, 1988; Thomson, 1988;
McKinney et al, 1990). These models
assume that any perturbation in development sets off a cascade of perturbations in
subsequent stages, and that later perturbations, by affecting fewer total subsequent
events, will have more limited phenotypic
effects. It has been suggested that early
embryogenesis ought therefore to be more
refractory to evolutionary change than later
stages of development (e.g., Gould and
Lewontin, 1979; Arthur, 1988; Buss, 1987;
Thomson, 1988; McKinney et al, 1990).
This is essentially a recasting of von Baer's
first law of development in terms of information theory.
This neo-von Baerian view predicts stagespecific biases in rates of evolution in developmental processes: the ontogenies of closely
related species should differ only in the later
developmental processes, while distantly
related species should show not only a
greater number of differences in later development, but also some modifications in earlier stages. Although many cases can be cited
to support these predictions, many other
RATES OF DEVELOPMENTAL EVOLUTION
examples clearly violate them. Familiar
modifications in early phases of development include larval adaptations for dispersal, feeding, and defense (Garstang, 1922;
Strathmann, 1978, 19886), and poecilogony, the presence of two developmental
modes in a single species (Giard, 1904;
Hoagland and Robertson, 1988). Evolutionary modifications in very early developmental processes can occur at the cellular
and molecular level, even among closely
related species (Elinson, 1987; Wray and
McClay, 1989; Jeffery and Swalla, 1990;
Wray and Raff, 1991a). The most dramatic
examples concern transformations in developmental mode, which involve extensive
modifications in mechanisms of early
development (Wray and Raff, 1991a), but
which have so little effect on adult morphology that species with planktotrophic and
lecithotrophic development are often classified as congeners (Mileikovsky, 1971).
Furthermore, the few estimates for r at
the beginning of development (oogenesis)
that are available (Table 1) are comparable
to estimates for r at the end of development
(morphogenesis of adult structures). To what
extent this can be generalized will only be
answered by the analysis of additional data.
The examples presented earlier in this paper
demonstrate, however, that early development does not necessarily evolve at slower
rates than late development.
Developmental constraints on the
origin of novel body plans?
The Cambrian "explosion" of body plans
is perhaps the single most striking feature
of the metazoan fossil record. The rapidity
with which phyla and classes appeared during the early Plaeozoic, coupled with much
lower rates of appearance for higher taxa
since, poses an outstanding problem in macroevolution. One explanation for this pattern is that developmental programs have
become too constrained by interaction since
the early radiation of metazoans to allow
the origin of new body plans (Reidel, 1978;
Erwin and Valentine, 1984; Levinton, 1988).
This hypothesis make a testable prediction: that the kinds of developmental differences that distinguish phyla and classes
should differ qualitatively from the differ-
131
ences that distinguish species and genera.
There is, however, empirical evidence to the
contrary. Some of the differences in developmental mechanism that have evolved
within the genus Heliocidaris are comparable to those that distinguish classes and
phyla: determinate vs. indeterminate cell
divisions, mode of gastrulation, and cell
cleavage pattern (Henry and Raff, 1990;
Wray and Raff, 1990, 1991a, b). Another
example is the "deuterostome gastropod"
Paludina, whose mouth develops at a secondary site rather than from the blastopore
(Verdonk and van den Biggelaar, 1983).
Developmental mechanisms characteristic of phyla and classes can at least occasionally change within families and genera.
Loss of evolutionary flexibility in developmental pathways does not therefore
appear to be a sufficient explanation for the
reduced rate of origins of new body plans
since the lower Paleozoic, although it may
provide a partial explanation in some
groups. Evolutionary changes in the fundamental processes of early development
may be more common at lower taxonomic
ranks than we realize: such changes often
are not morphologically obvious, and the
phylogenetic range of species whose embryos
have been experimentally manipulated
remains limited.
An interesting question that emerges from
these considerations is why some aspects of
development are so conservative when they
clearly can change. Answering this question
will require an integrated approach to
studying the cellular and molecular bases
for evolutionary transformations in development (Wray and Raff, 1991a), along with
a clearer understanding of the functional
implications of these transformations
(Strathmann, 19886; Emlet, 1991).
CONCLUSIONS
This paper presents some estimates of
evolutionary rates in developmental processes. The data set is limited, but some
general trends are probably significant. With
regard to tempo, developmental processes
can evolve at approximately the same rate
as adult morphology. And with regard to
mode, rates ofdevelopmental evolution can
vary by 10- or 100-fold within a clade or
132
GREGORY A. WRAY
lineage. The extent to which these trends
are characteristic of metazoan developmental evolution in general cannot be determined until considerably more data are analyzed. Rates of developmental evolution can
be calculated in any group where comparative developmental data, a robust phylogenetic hypothesis, and good estimates for
divergence times are available. Additional
analyses of rates of developmental evolution are feasible and are likely to provide
important insights into macroevolutionary
phenomena.
ACKNOWLEDGMENTS
Richard Strathmann, Jon Havenhand,
and Sherryl Broverman provided many useful suggestions and criticisms during the
ontogeny of this manuscript. This research
was supported by NIH postdoctoral fellowship GM12495.
APPENDIX
Calculating minimum proportional rates of evolution
when the ancestral condition must be inferred.
We wish to calculate the minimum average proportional rate of evolution, r, in a quantitative developmental parameter, P. To do this, we must reconstruct
the value of P in the ancestor of two extant species,
where P can be directly measured. The most conservative reconstruction of P in the ancestor (Panc) is as
the geometric mean (proportional mean) of P measured
in the two extant species (P, and P2). The geometric
mean will yield equal values of opposite sign r for the
transformations Panc - P, and ParK -> P2 (see Fig. 2B).
To demonstrate that this is so, set these transformations (from equation (1) in the main text) as equal, and
solve for Panc:
ln(P.nc) -
ln(P,J - ln(P2)
where td is the divergence time. This reduces to:
1n(P • P )
P™ = e~r~
(3)
which is the geometric mean of P, and P2. Substituting
P.nc back into equation (1) and simplifying gives equation (2) of the main text, a general formula for calculating minimum average proportional rates of change
based upon differences in extant species.
REFERENCES
Arthur, W. 1988. A theory of the evolution of development. Wiley, Chichester.
Baverstock, P. R. and M. Adams. 1987. Comparative
rates of molecular, chromosomal and morphological evolution in some Australian vertebrates. In
K. S. W. Campbell and M. F. Day (eds.), Rates of
evolution, pp. 175-188. Allen & Unwin, London.
Buss, L. 1987. The evolution of individuality. Princeton Univ. Press, Princeton, New Jersey.
Campbell, K. S. W. and M. F. Day. 1987. Rates of
evolution. Allen & Unwin, London.
Cherry, L. M., S. M. Case, and A. C. Wilson. 1978.
Frog perspective on the morphological difference
between human and chimpanzee. Science 200:209211.
Daugherty, C. H., F. W. Allendorf, W. M. Dunlap, and
K. L. Knudsen. 1983. Systematic implications
of geographic patterns of genetic variation in the
genus Dicamptodon. Copeia 1983:679-691.
Donoghue, M. J. 1989. Phylogenies and the analysis
of evolutionary sequences, with examples from
seed plants. Evolution 43:1137-1156.
Eldredge, N. and S. J. Gould. 1972. Punctuated equilibria: An alternative to phyletic gradualism. In T.
J. M. Schopf (ed.), Models in paleobiology, pp. 82115. Freeman, Cooper & Co., San Francisco.
Elinson, R. P. 1987. Changes in developmental patterns: Embryos of amphibians with large eggs. In
R. A. Raff and E. C. Raff (eds.), Development as
an evolutionary process, pp. 1-21. Liss, New York.
Emlet, R. B. 1988. Larval form and metamorphosis
of a "primitive" sea urchin, Eucidaris thouarsi
(Echinodermata: Echinoidea: Cidaroida), with
implications for developmental and phylogenetic
studies. Biol. Bull. 174:4-19.
Emlet, R. B. 1991. Functional constraints on the evolution of larval forms of marine invertebrates:
Experimental and comparative evidence. Amer.
Zool. 31:707-725.
Emlet, R. B., L. R. McEdward, and R. R. Strathmann.
1987. Echinoderm larval ecology viewed from the
egg. In M. Jangoux and J. Lawrence (eds.), Echinoderm studies 2, pp. 55-136. Balkema Press,
Amsterdam.
Erwin, D. H. and J. W. Valentine. 1984. "Hopeful
monsters," transposons, and metazoan radiation.
Proc. Nat. Acad. Sci. U.S.A. 81:5482-5483.
Farris, J. S. 1982. Outgroups and parsimony. Syst.
Zool. 31:328-334.
Foote, M. and R. H. Cowie. 1988. Developmental
buffering as a mechanism for stasis: Evidence from
the pulmonate Theba pisana. Evolution 42:396399.
Garstang, W. J. 1922. The theory of recapitulation:
A critical re-statement of the biogenetic law. Zool.
J. Linn. Soc. 35:81-101.
Giard, A. 1904. La Poecilogenie. Sixieme Congress
Internationale de Zoologie, Compte Rendue des
Seances 6:617-646.
Gingerich, P. D. 1983. Rates of evolution: Effects of
time and temporal scaling. Science 222:159-161.
Gould, S. J. and R. C. Lewontin. 1979. The spandrels
of San Marco and the Panglossian paradigm: A
critique of the adaptationist paradigm. Proc. R.
Soc. London B 205:581-598.
Haldane, J. B. S. 1949. Suggestions as to quantitative
measurement of rates of evolution. Evolution 3:
51-56.
Hallam,A. 1975. Evolutionary-size increase and Ion-
RATES OF DEVELOPMENTAL EVOLUTION
gevity in Jurassic bivalves and ammonites. Nature
258:493^*96.
Hallam, A. 1978. How rare is phyletic gradualism,
and what is its evolutionary significance? Evidence
from Jurassic bivalves. Paleobiology 4:16-25.
Hansen, T. A. 1982. Modes of larval development
in Early Tertiary neogastropods. Paleobiology 8:
367-377.
Henry, J. J. and R. A. Raff. 1990. Evolutionary
changes in the mechanisms of dorsoventral fate
determination in the direct developing sea urchin
Heliocidaris erythrogramma. Dev. Biol. 141:55—
69.
Hoagland, K. E. and R. Robertson. 1988. An assessment of poecilogeny in marine invertebrates: Phenomenon or fantasy? Biol. Bull. 174:109-125.
Jablonski, D. 1986. Larval ecology and macroevolution in marine invertebrates. Bull. Mar. Sci. 39:
565-687.
Jagersten, G. 1972. Evolution of the metazoan life
cycle: A comprehensive theory. Academic Press,
New York.
Jeffery, W. R. and B. J. Swalla. 1990. Anural development in ascidians: Evolutionary modification
and elimination of the tadpole larva. Sem. Dev.
Biol. 1:253-261.
Katz, M. J. 1983. Ontogenetic mechanisms: The
middle ground of evolution. In J. T. Bonner (ed.),
Evolution and development, pp. 207-212. SpringerVerlag, Berlin.
Keigwin, L. D. 1978. Pliocene closing of the Isthmus
of Panama, based on biostratigraphic evidence
from nearby Pacific Ocean and Caribbean Sea
cores. Geology 6:630-643.
Keigwin, L. D. 1982. Isotopic paleoceanography of
the Caribbean and east Pacific: Role of the Panama
uplift in late Neogene time. Science 217:350-353.
Kier, P. M. 1984. Echinoids from the Triassic (St.
Cassian) of Italy, their lantern supports, and a
revised phylogeny of Triassic echinoids. Smith.
Contr. Paleobiol. 56:1-41.
Kume, M. and K. Dan. 1968. Invertebrate embryology. Transl. J. C. Dan. NOLIT Publ. House,
Belgrade.
Larson, A. 1984. Neontological inferences of evolutionary pattern and process in the salamander
family Plethodontidae. Evol. Biol. 17:119-217.
Lessios, H. A. 1981. Divergence in allopatry: Molecular and morphological differentiation between sea
urchins separated by the Isthmus of Panama. Evolution 35:618-634.
Lessios, H. A. 1990. Adaptation and phylogeny as
determinants of egg size in echinoderms from the
two sides of the Isthmus of Panama. Amer. Nat.
135:1-13.
Levinton, J. 1988. Genetics, paleontology and macroevolution. Cambridge Univ. Press, Cambridge.
MacFadden, B. J. 1985. Patterns of phylogeny and
rates of evolution in fossil horses: Hipparions from
the Miocene and Pliocene of North America.
Paleobiology 11:245-257.
Maddison, W. P., M. J. Donoghue, and D. R. Maddison. 1984. Outgroup analysis and parsimony.
Syst. Zool. 33:83-103.
133
Maynard Smith, J., R. Burian, S. Kauffman, P. Alberch,
J. Campbell, B. Goodwin, R. Lande, D. Raup, and
L. Wolpert. 1985. Developmental constraints and
evolution. Quart. Rev. Biol. 60:265-287.
McKinney, M. L., K. J. McNamara, and L'. G. Zachos.
1990. Heterochronic hierarchies: Application and
theory in evolution. J. Hist. Biol. 3:269-287.
McMillan, W. O., R. A. Raff, and S. R. Palumbi. 1992.
The evolutionary and population genetic consequences of developmental changes resulting in low
dispersal of sea urchins. Evolution. (In press)
McNamara, K. J. 1988. The abundance of heterochrony in the fossil record. In M. L. McKinney
(ed.), Heterochrony in evolution, pp. 187-325. Plenum, New York.
Mileikovsky, S. 1971. Types of larval development
in marine bottom invertebrates, their distribution
and ecological significance: A re-evaluation. Mar.
Biol. 10:193-213.
Needham, J. 1933. On the dissociability of developmental processes. Biol. Rev. 8:180-223.
Nijhout, H. F. 1990. Metaphors and the role of genes
in development. BioEssays 12:441-446.
Okazaki, K. 1975. Normal development to metamorphosis. In G. Czihak (ed.), The sea urchin
embryo, pp. 177-232. Springer-Verlag, Berlin.
Raff, R. A. 1987. Constraint, flexibility, and phylogenetic history in the evolution of direct development in sea urchins. Dev. Biol. 119:6-19.
Raff, R. A., J. A. Anstrom, C. J. Huffman, D. S. Leaf,
J. H. Loo, R. M. Showman, and D. E. Wells. 1984.
Origin of a gene regulatory mechanism in the evolution of echinoderms. Nature 310:312-314.
Raff, R. A. and G. A. Wray. 1989. Heterochrony:
Developmental mechanisms and evolutionary
results. J. Evol. Biol. 2:409^34.
Raff,R.A.,J.J.Henry,andG.A.Wray. 1991. Implications of radical changes in early development
for concepts of developmental constraint. In L.
Warren and H. Koprowski (eds.), Perspectives in
evolution, 189-207. A. R. Liss, New York. (In
press)
Reidel, R. 1978. Order in living organisms: A systems
analysis of evolution. Transl. R. P. S. Jefferies.
Wiley, Chichester, U.K.
Schroeder, T. 1981. Development of a "primitive"
sea urchin (Eucidaris tribuloides): Irregularities in
the hyaline layer, micromeres, and primary mesenchyme. Biol. Bull. 161:141-151.
Simpson, G. G. 1944. Tempo and mode in evolution.
Columbia Univ. Press, New York.
Simpson,G.G. 1953. The majorfeatures ofevolution.
Columbia Univ. Press, New York.
Smith, A. B. 1984. Echinoid palaeobiology. Allen &
Unwin, London.
Smith, A. B. 1988a. Fossil evidence for the relationships of extant echinoderm classes and their times
of divergence. In C. R. C. Paul and A. B. Smith
(eds.), Echinoderm phylogeny and evolutionary
biology, pp. 85-97. Oxford Univ. Press.
Smith, A. B. 19886. Phylogenetic relationship, divergence times, and rates of molecular evolution for
camarodont sea urchins. Mol. Biol. Evol. 5:345356.
134
GREGORY A. WRAY
Stanley, S. M. 1979. Macroevolution: Pattern and
process. Freeman, San Francisco.
Stanley, S. M. 1985. Rates of evolution. Paleobiology
11:13-26.
Stanley, S. M. and X. Yang. 1987. Approximate stasis
for bivalve morphology over millions of years: A
multivariate, multilineage study. Paleobiology 13:
113-139.
Strathmann, M. F. 1987. Reproduction and development of marine invertebrates of the northern
Pacific coast. Univ. Washington Press, Seattle.
Strathmann, R. R. 1978. The evolution and loss of
feeding stages of marine invertebrates. Evolution
32:894-906.
Strathmann, R. R. 1985. Feeding and non-feeding
larval development and life history evolution in
marine invertebrates. Ann. Rev. Ecol. Syst. 16:
339-361.
Strathmann, R. R. 1988a. Larvae, phylogeny, and
von Baer's law. In C. R. C. Paul and A. B. Smith
(eds.), Echinoderm phylogeny and evolutionary
biology, pp. 53-68. Oxford Univ. Press, London.
Strathmann, R. R. 1988fc. Functional requirements
and the evolution of developmental patterns. In
R. D. Burke, P. V. Mladenov, P. Lambert, and R.
Parsley (eds.), Echinoderm biology, pp. 55-61.
Balkema, Amsterdam.
Swofford, D.L. and W. P. Maddison. 1987. Reconstructing ancestral states under Wagner parsimony. Math. Biosci. 87:199-229.
Thomson, K. S. 1988. Morphogenesis and evolution.
Oxford Univ. Press, London.
VanValen, L. 1974. Two modes of evolution. Nature
252:298-300.
Verdonk, N. H. and J. A. M. van den Biggelaar. 1983.
Early development and the formation of the germ
layers. In N. H. Verdonk and J. A. M. van den
Biggelaar (eds.), The Mollusca, Vol. 3, pp. 91-122.
Academic Press, New York.
Waddington, C. H. 1957. The strategy of the genes.
Allen & Unwin, London.
Wray, G. A. and D. R. McClay. 1988. The origin of
spicule-forming cells in a "primitive" sea urchin
(Eucidaris tribuloides) which appears to lack primary mesenchyme cells. Development 103:305315.
Wray, G. A. and D. R. McClay. 1989. Molecular
heterochronies and heterotopies in early echinoid
development. Evolution 42:803-813.
Wray, G. A. and R. A. Raff. 1990. Novel origins of
lineage founder cells in the direct-developing sea
urchin Heliocidaris erythrogramma. Dev. Biol. 141:
41-54.
Wray, G. A. and R. A. Raff. 1991a. The evolution
of developmental strategy in marine invertebrates.
Trends Ecol. & Evol. 6:45-50.
Wray, G. A. and R. A. Raff. 1991ft. Rapid evolution
of gastrulation mechanisms in a sea urchin with
lecithotrophic larvae. Evolution 45:1741-1750.
Wright, S. 1980. Genie and organismic selection.
Evolution 34:825-843.