ontogeny of descendants.
This definition is simple
Dissociation and Heterochrony
In 1770, Daines Barrington an English jurist and
general scholar, published a paper in the Philosophical
Transactions of the Royal Society”’ entitled: ‘Account
of a very remarkable young musician.’ He wrote of the
visit to England, six years previously, of a prodigy
named Johannes Chrysostomus Wolfgangus Theophilus
Mozart. (Remember that Mozart, during his English
visit, was an untested and puzzling eight year old
curiosity, not yet a principal icon of human achieve-
ment.)
Two aspects of young Mozart, united by the single
theme of dissociability, particularly intrigued Barr-
ington. First, he marvelled that musical talent could be
so mature and refined in a young boy, otherwise so
typically childish. The musical faculty must be a
separable component of mental ability. Secondly. he
asked Mozart to sit at the harpsichord and improvise
compositions expressing the single emotions of love and
rage. Mozart’s immediate and consummate success
convinced Barrington that the whirligig of human
emotional life must be dissectable into separate
essences.
This theme of dissociation into separate modules,
each subject to largely independent manipulation and
alteration, lies at the heart of any hope for resolving the
natural development (ontogeny or phylogeny) of any
complex system, biological or otherwise. Indeed, the
first great argument against the possibility of evolution
- Cuvier’s correlation of parts - was rooted in a denial
of dissociability. In his Discours prdliminairei2) of 1812,
Cuvier portrayed animals as so integrally connected
that a change in one part, however slight or subtle,
would necessarily require compensatory alteration in
every other feature in order to maintain a functioning
creature. Since such wholesale restructuring for each
change is inconceivable, evolution becomes an impossi-
bility. Cuvier’s argument is impeccable in logic and
would, indeed, debar evolution if its premise of
absolute integration were valid.
Heterochrony achieves its special importance in
studies of development and evolution because it
embodies this central enabling theme of modularity and
dissociation. Heterochrony is defined as phyletic
change in the timing of development, such that features
of ancestors shift to earlier or later stages in the
enough, but gives rise to a host of practical and
theoretical difficulties, and to a rash of consequent
literature. Consider just two issues. First, a shift in
timing relative to what? Newtonian time in days,
developmental time in stages? And what is to be done
with paleontological material when no absolute times
can be ascertained? Can a surrogate like body size then
be used? 1 tricd, with limited success, to develop a
‘clock model’ for resolving these issues (ref. 3, pp. 246-
262), but the later formalization of Alberch et al.i4) is
preferable and has generally been accepted as an
‘industry standard.‘
Second, a subject can easily be destroyed by
attempting to encompass too much. In one sense, and
almost by necessary logic, all change must involve
timing. If heterochrony involves any change describable
by speeds of reactions, then we might as well retire the
concept as a synonym of evolution. Heterochrony is a
principle of historical legacy and its later, creative
conscquences - evolution by relative temporal shift of
features already present in ancestors, as opposed to
introductions of novelties. (I am well aware that
apparent novelty, in morphological or functional
aspects, may be constructed by altering the rates of
ancestral developmental sequences - a fact that
demands further complexity in definition. See
Alberch@)for an cxcellent discussion and resolution of
this issue.) In any event, heterochrony, which once
languished in the deepest doghouse of neglected
concepts. now ranks among the most popular of
evolutionar topics (see, for instance, recent books by
McKinney(4 and McKinney and M~Namara(~)) and
holds greatest promise as a unifier of the cardinal
subjects of ontogeny and phylogeny.
A Short History: Haeckel’s Legacy and
Darwinism’s Neglect
The study of development should be the unifying point
for integrating reductionist and mechanical studies of
genes and cells with historical and narrative accounts of
life’s phylogeny, in short, the tocus for a truly
encompassing biology. Aristotle granted embryology
this central role and Darwin’s grandfather Erasmus
developed an ontogenetic theory of phylogeny (as were
most evolutionary accounts in the progressivist climate
of the Enlightenment). and even expressed it in heroic
couplets:
See, without parents, by spontaneous birth
Rise the first specks of animated earth.. .
In Charles Darwin’s age, of course, Haeckel’s
biogenetic law, or theory of recapitulation, provided
the canonical account of relationship between ontogeny
and phylogeny. Expressed in heterochronic terms,
Haeckel’s principle required a virtually universal
acceleration, or speeding up of devclopment, through
evolutionary sequences. In this way, adult characters of
ancesters could become juvenile features of descend-
ants and, ultimately, ontogeny might be read as an
epitome of phylogeny: the earlier the stage of
development, the longer ago it served as an ancestral
adult form.
Haeckel’s theory had a profound influence across
Western culture (see chapter 5 in G ~ u l d ( ~ probably
)),
rivalling that of natural selection itself. Recapitulation
directed primary school curricula of the late 19th
century (since young children were like ‘primitive’
adults, and the legends of these ancient times would
therefore resonate with their juvenile souls); it
buttressed racism and its colonialist extensions (since
‘lower‘ people were like juveniles of ‘higher’ races and
needed the same ‘protection’, read domination, as
children); and it provided a basis for Freud’s psycho-
logical theories (the Oedipus complex repeats, in
juvenile males, an actual event of parricide once
inflicted by adult sons upon a ruling father - see his
books Totem and Taboo, and Moses and Monotheism).
Haeckel‘s doctrine suffered a spectacular collapse in
late 19th and early 20th century biology, leading to an
overreaction and a virtual suppression of the large (and
always vital) subject of relationships between ontogeny
and phylogeny - a classic illustration of the old clichC
(particularly relevant here, given the subject matter), of
throwing out the baby with the bathwater. Two primary
reasons underlay Haeckel’s fall and the suppression of
evolutionary interest in ontogeny:
1. Temper of the times. Late 19th century biology,
especially in Germany (then the center of advanced
research), favored a rigorously mechanistic and exper-
imental approach embodied in Roux‘s Entwicklungs-
mechanik for studies of development. Roux’s school
viewed the inferential reconstruction of family trees
from evidence of ontogeny as, at best, of limited value
and merely descriptive in nature and, at worst, as
fatuous. Roux himself wrote (ref. 8, p. 253): ‘To be
sure, we agree with Haeckel’s dictum ... but the
biogenetic law merely designates the fact of repetition
... It teaches us nothing about the operating causes and
their intensities. The cxperimental study of develop-
mental mechanics can bring us this knowledge.’ Less
diplomatically, the great American disciple E. B.
Wilson spoke for modernism from the newly-founded
Marine Biological Laboratory at Woods Hole (ref. 9,
pp. 103-104): ‘The search after suggestive working
hypotheses in embryological morphology has too often
led to a wild speculation unworthy of the name of
science; and it would be small wonder if the modern
student, especially after a training in the methods of
more exact sciences, should regard the whole phylogen-
etic aspect of morphology as a kind of speculative
pedantry unworthy of serious attention.’
2. Logic or argument. Recapitulation’s need for
universal acceleration found its easiest justification
under Lamarckian models of inheritance because new
features acquired in adult life are added to the linear
sequence of inherited ontogeny, thereby pushing older
features back. Darwinian recapitulationists, like
August Weismann, were forced into an uncomfortable
argument: we don’t know how heredity works, but
when we find out, its mechanics will include a
justification for universal acceleration.
Mendelism, reborn in the early 1900s’ thoroughly
thwarted this expectation. If inheritance is particulate
in a Darwinian world; if genes make enzymes and
enzymes control the rates of processes; and if natural
selection might favor altered rates in any direction
(depending on circumstances); then why should univer-
sal acceleration (or any universal trend) take place?
Phyletic slowing of developmental rates should be as
orthodox in evolution as Haeckel’s speeding up. And,
with no genetic basis for universal (or even preferred)
acceleration. the rationale for recapitulation disappears
forthwith.
This Mendelian reform should have ushered in a new
life and heyday for study of relationships between
ontogeny and phylogeny. The old Haeckelian strangle-
hold had been broken and a full taxonomy of
heterochronies could now be elaborated (slowing down
and speeding up of various modules with respect to each
other and to ancestral modules) - and their evolution-
ary effects illustrated and evaluated. (Indeed, this
promise has been fulfilled during the past 20 years and
does now underlie the current fruitfulness and excite-
ment of the topic). The chief paradox of this subject in
our century lies in the failure of this renaissance to
occur, and in the occurrence of its opposite, the descent
of a pall of inattention (if not downright hostility) over
the topic until recently. I see two major reasons for thk
paradoxical neglect:
1. Sociology. The unhappiness of experimentalists
at speculative excesses in Haeckelian ph ylogenizing,
and the sharp disappearance of Haeckel’s needed
rationale with the rise of Mendelism, gave the entire
subject such a bad odor that most sober scientists simply
stayed away, even though other aspects of the field
cried for attention. I will never forget the opening line
of harsh criticism poured by a fine biologist upon a
colleague who dared to raise thc subject in 1971, as 1
began research for Ontogeny and Phylogeny: ‘There are
still those who would Haeckel biology’ (quoted in
~ o u i d (p.
~ )2).
2. Theory. Nothing is quite so frustrating as being
ignored; one would rather be explicitly reviled (and
thereby taken seriously). Embryological issues received
short shrift under the strict Darwinism that came to
prevail in evolutionary circles after the movement
known as the ‘Modern Synthesis’took root in the 1930’s
and spread to orthodoxy through its time of greatest
triumph in the Darwinian centennial celebrations of
1959. Nothing in the strict Darwinian paradigm suggests
hostility to developmental issues, but the theory’s
central logic offers precious little space for the major
internalist and structuralist themes of embryology
either. Darwinism is fundamentally an externalist
theory of adaptation, step by insensible step, to
changing local environments. Internal genetic factors
for development supply the copious, small-scale,
fundamentally random variation that acts only as rah
material in Darwin’s system. The direction and scope of
change are externally determined by natural selection
based on selective pressures of surrounding environ-
ments.
Why would anyone grant much attention to develop-
ment in such a system? Development produces
variation, but variation itself plays no directing role in
its curiously formless status as an isotropic sphere
around a modal morphology. In effect, variation may
be taken simply as a ‘given,’ as a ‘primitive term’ in the
argument. (Darwin himself had no other choice, given
the ignorance of his age about causes of variation and
modes of inheritance.) If variation is always present,
and usually unconstraining, why worry unduly about its
sources or consequences, so long as one can be
confident about its availability. Moreover, Darwinism
also downplays the essential internalist theme of small
change amplified through development to large effect
on adult phenotypes - for strict Darwinians have always
favored accumulative models of microevolutionary
extrapolation for major changes, that is. step by
countless step, upward from the world of insensible
increments where development becomes largely irrel-
evant.
The Modern Synthesis, Darwinism’s triumph, was
integrative in two senses: first. in bringing Darwinism
and Mendelism together: second, in gathering the
classical subjects of biology under its umbrella, one by
one - Mayr for systematics, Simpson for paleontology,
Stebbins for botany, White for cytology, etc. However,
embryology and development stayed outside the
penumbra. De Beer’s short book Embryos and
Ancestors (1940)(1°),was enjoyed but unexploited while
Goldschmidt ( 1940)(11), who touted the classical
internalist themes of constraint and large-scale change,
was reviled. Waddington, the embryologist (and ex-
paleontologist) who moved most easily in synthetic
circles, would blurt out in frustration that his colleagues
devised elegant models of allelic change in populations,
or studied the most minute shifts of microevolutionary
phenotypes, while he wanted to know how evolution
made ’lions and tigers and things.’ Development stood
very low on the agenda. if not beyond the pale, of the
synthetic theory.
The New Potential Union of Evolution and
Development, and the Centrality of Heterochrony
Nothing in my entire career has brought me more
pleasure than watching (largely passively, after provid-
ing a little push by publishing Ontogeny and Phylogeny
in 1977) the reintegration of evolution and develop-
ment, from some cautious forays on heterochronic
themes to a virtual cascade today as we finally begin to
unlock the genetic basis of development. I see two
major reasons for this rerouting in parallel of central
themes that had diverged for bad reasons.
1. Elimination uf theoretical harriers. Strict Dar-
winism, as argued above, contains little room for the
internalist themes best illustrated by development:
constraints and substantial changes. While strict
Darwinism prevailed, development would remain in
limbo. However, the hegemony of strict Darwinism has
been broken in the past twenty years. This has occurred
in a fruitful way that leaves the Darwinian mechanism
intact as a centerpiece. while expanding the realm or
evolutionary forces and levels to include much else that
is centrally concerned with development. The reasons
for this expansion are many and complex, and far
beyond the scope of this article. However, they include.
(i) Expansion of levels above and below Darwinism’s
nearly exclusive focus on individual bodies and their
struggle for reproductive success - from alleles or
nucleotides as the focal level in Kimura’s(12)theory ot
neutralism, to species in the theory of punctuated
equilibrium(13), to clades in theories of mass extinc-
tion(14); (ii) Expansion of causes of change to include
chance at all levels (allelic neutrality to mass extinction)
and developmental constraints as positive channels of
preferred change, not merely as negative forces,
preventing advantageous alteration.
Much of this ferment can be captured in the
beautifully apt metaphor first roposed by Darwin’s
P
brilliant cousin Francis Galton( 5, in 1S69, and widely
invoked by evolutionists (including Bateson and de
Vries) before development slipped from popularity.
Galton argued that we should view organisms as
polyhedral solids. They may not budge unless pushed
by the external force of natural selection but, when
shoved, they can only move in limited directions
specified by internal features of geometry (built, of
course, by development). This internal architecture
embodies both great development themes: constraint
(the polyhedron can only move from one facet to
another, and longer pathways of change are internally
specified by the forms and interrelations of facets), and
substantial alteration (movement must occur by a flip
from one stable facet to another). By contrast, in the
world of strict Darwinism, organisms are spheres and
they move wherever the pool cue of natural selection
and thc table of environment specify - without any
‘pushing back’ from an internal geometry constructed
developmentally.
So profound is this shift of concern that ‘developmen-
tal constraint’ has become one of the hottest topics in
evolution(lb-ls),so much so that it threatens to become
a ‘buzzword’, too loosely defined and too encompass-
ing. Heterochrony has pride of place among develop-
mental themes in evolution primarily because it
represents the most fruitful ontogenetic approach to
both major internalist themes of constraint and
structurally-aided wbstantial change. Of all constraints,
standard sequences of ontogenetic change are the most
immediate and powerful on the obvious principle that
nature works more easily by rearranging what it has
than by constructing novelty. Each organism has a great
panoply of evolutionary change available within its own
growth because so many features alter allometrically
through growth, often producing substantial modifi-
cations of ontogeny. Given the principle of modularity
(see introductory section), these sequences can be
dissociated and the relative timings altered. Hetero-
chrony is defined as temporal alteration plus dis-
sociation and it is therefore the chief focus for using
current pathways (constraints) in evolutionary change.
Similarly, heterochrony has long supplied the pri-
mary macroevolutionary theme for ontogenetic con-
cepts, namely the proposal that small genetic changes,
acting early in development, can cascade to large
phenotypic effects, providing thereby one of the few
legitimate Darwinian rationales for rapid, large-scale
change. Such potential pathways are almost always
heterochronic, as in the putative origin of higher taxa by
progenesis (or the early maturation of larvae with
interesting, and creative, mixtures of juvenile and adult
characters, given modularity and the linkage of some
features, but not others, to accelerated maturation).
2. Reduction of practical barriers. We (evolutionists
like myself, that is) have contributed by expanding a
theory to accommodate developmental perspectives
but you (developmental biologists who read this essay)
have made a far more important contribution to our
remarriage by devising the techniques that makc
development tractable. Although I grant this theme
relatively little space - because, as an outsider looking
in, 1 understand it so poorly, and would be foolish to
pontificate within an issue filled with articles by real
experts -this growing tractability formed my reason for
writing this article in the first place.
My frustration was immense when I published
Ontogeny and Phylogeny in 1977. I sensed that
development was vital to any fuller and more adequate
evolutionary theory, and I recognized that hetero-
chrony would form a practical focus for study. I wrote in
the introduction: ’This book is primarily a long
argument for the evolutionary importance of hetero-
chrony - changes in the relative timing of appearance
and rate of development for characters already present
in ancestors.’ But, when I came to examine the causes
of heterochronic change (rather than describing their
effects), I could find almost nothing, for knowledge of
the genetics of development was then a virtual blank. I
felt as though I stood far away watching the surface of a
growing ball, without any understanding of what made
it expand from inside and what produced its surface
architecture (not to mention its deep structure). All 1
could do was write a slim final chapter with a few brave
words (superficial, however well intentioned) on the
putative difference between structural and regulatory
genes. I even began the book with a line of apology
(1977, p. 1): ‘I am aware that I trcat a subject currently
unpopular. ’
All this has changed dramatically, thanks to your
work. Our understanding is still in early infancy, but we
now have the tools (from rapid and routine genetic
sequencing to labelling and tracing of gene products in
ontogeny) to unlock the genetics of development. From
an immediately heterochronic perspective, specific
genes that regulate rates of developrnent have been
identified and characterized in Caenorhabditis eleguns
(wherc they have been called ‘heterochronic genes’)
and other creatures(”).
More generally, we are beginning to glimpse the
overall development of some complex creatures as a
hierarchical series of increasing specificity: from, in
Drosophila for example, basic gradient makers (like
b~coid)to segmentation genes (like fushi tarazm) to
differentiators of particular segments (the homeotics),
And the developing explanation of homeotic function,
with discontinuous phenotypic effects, arising along a
gradient of concentration, produced by genes arranged
in the same order, is among the most elegant stories in
all of biology.
Moreover. the clear genetic homology (using the
word in its true evolutionary sense of similarity by
common descent) between the Drosophila HOM
complex and the fourfold-repeated HOX sequences of
mice and other vertebrates shows that these disparate
phyla still carry complex legacies of common ancestry.
(And revives biology’s oldest dream of unification,
Geoffroy’s early 19th century claim for a common
architecture to all animal life, as Goethe had advanced
a similar theory for plants, based on a leaf archetype.)
The issue of morphological homology is still open for
homologous genes may commandeer different pheno-
typic apparatuses, and t h e final morphological products
need not be homologues. For example, somites and
metameres are too structurally and developmentally
different to qualify, and this has led to the traditional
denial of homology between the phyla. But some
remarkable similarities have recently been found
between arthropod metameres and a second system of
vertebrate segmentation centered on rhombomeres of
the developing mid and hindbrain, branchial arches and
some cranial nerves. Hunt et a1.(20)have just published
the intriguing discovery that paralogues in the four
vertebrate HOX complexes show identical expression
limits in the rhombomeres. cranial ganglia and
branchial arches, but do not work in such a coordinated
way in the trunk somites. (As a pure speculation, may
this not indicate a phyletically primitive status for the
rhombomeric system, with the somitic system sur-
added, as the HOX duplications continue to build
rhombomeres on the old pattern, while working
differently with the unique and later somites)? If a
paleontological word might be in order, early agnathans
(and some pre-vertebrate chordates) have very promi-
nent branchial baskets, presumably a filtration device
by original function. with a smaller somitic system, as if
a ininor trunk and tail were added to a massive frontal
device. Rhombomeres are now transient in vertebrates
(as are branchial arches in tetrapods), but current
prominence is no mark of phyletic precedence. (The
frontal segmentation is phylefically confused by the
formation of skeletal structures in the branchial region
from neural crest, a later and clearly novel vertebrate
feature, apparently not influenced by HOX genes.
Developmental biologists who wish to explore the
possible morphological homology of arthropod meta-
meres and vertebral frontal segmentation might try to
subtract the neural crest derivatives and focus on
development of what may remain from the old
branchial basket segmentation.)
In any case, this developing (and fast-breaking) story
strongly increases the probable significance of con-
straint and heterochrony in evolution and greatly
diminishes the fading, strict Darwinian idea that
evolution may be traced as insensibly transitional
adaptation to external environments, without close
attention to how internal architecture channels and
structures the potential pathways of substantial change.
If groups so disparate as arthropods and vertebrates
bear substantial homology in basic morphological
design, then constraints on pathways arc legion and
life’s commonality will be resolved more by the study of
development than by adaptation.
A Mozartian Epilogue: Cosi Fan Tutte
To end where I began (as I write this article in the very
month of Mozart’s bicentennial), consider his opera
Cosi Fan Tutte as an epitome of the history of
relationships between developmental and evolutionary
biology. The two couples of the opera are initially
joined in tirades of romantic pledges about eternal
devotion, but in total youthful naivete (and therefore
with no firm foundation) - just as ontogeny and
phylogeny were so strongly fused under the false
banner of Haeckel’s biogenetic law. The couples are
then separated in growing hostility and claims of
betrayal - only to reunite at the end when they accept
each other’s errors, foibles and failings, thereby forging
a true bond based on practical knowledge rather than
inconstant emotion (a favorite theme of the Age of
Reason, quite explicitly extolled by Mozart against
romanticism). So, too, for development and evolution,
two fields that needed to separate when the false focus
of previous union became clear, but that now can
reunite, both older and wiser, with the discovery of a
true and lasting basis for synthesis. It is an old story -
cosifan tutte (all women are like that - men too, as the
non-sexist opera makes quite clear). Professions, in
many ways, are like people. And when they finally join
for the right reason, look for the cascade of fruitful
progeny.
References
I Barrington, D. (1770). Account of a very remarkable young musician. Phil.
Trans. Roy. Soc. London 60, 54-64.
2 Cuvier, G . (1812). Discours pr4limiminaire, in Rechercher sur 1cs ossemens
fossiles. Paris: Dcterville.
3 Gould, S. J. (1977). Onmgeny and Phylogeny. Cambridge, MA: Harvard
Universily Press
4 Alberch, P., Gould, S. J., Oster, C. F. and Wake, D. B. (1979). Size and
shape in ontogeny and phylogeny. Paleobiology 5. 296.317.
5 Alherch, P. (1985). Problems with the interpretation of developmental
sequences. .Syst. Zoo/. 34, 46-58.
6 McKinney, M. J>., editor (1988). Heterochrony in Evohtion. Kew York:
Plenum Press.
7 McKinney, M. L. and McNamara, K. J. (1991). Heterorlzrony. The Ei:olution
of Onrogeny. New York: Plenum Press.
8 Roux, W. (1W5). Die ~ntwicklungsmcchanili. Eiii neurr Zweig der
biologischen Wissenschaft. VortrPge und Aufsiitze der EntwicklunRsmcchanili
dcr Organismen no. 1. Leipzig: W. Engelmann.
9 Wilson, E. B. (1894). The embryological criterion of homology. Uiol. Lect.
Mor. B i d . Lab. Woods Hole, pp. 101-124.
10 De Beer, G. K. (1940). Eruhryos urid ancectors. Oxford: Clarendon Press.
11 Goldschmidt, R. R. (1930). TIzr hfatcrinl Basis of Evolurion. New Haven,
CT: Yale University Presa.
12 Kimura, Motoo (1991). Neutral Evolution. In: S. 0 w w a and T. Hoiijo. eds.,
Evohrrion of Lf e . Tokyo: Springer-Verlag.
13 Eldredge, N. and Gould, S. J. (1972). Punctdated equilibria: An alternative
to pliyledc gradualism. In: ’I’ J..M. Schopf, ed.. Models in Paleobiology. San
Francisco: Freeman. Cooper & Company.
14 Alvarez, L. W., Alvarez, W., Asaro, F. and Michel. H. V. (1980).
Extraterrestrial cause for the Cretaceous-Terliary extinctioI1. Science 208. 1095-
1108.
15 Galton, F. (1869). Hereditary Genius. London.
16 Stearns, S. C. (1986). Natural selection and fitness, adaptation and
constraint. In: D. M. Raup and D. Jablonski, eds.. Patterns and Processes in the
History of Life. Berlin: Springer-Verlag.
17 Maynard Smith, J., Burian, J., Kauffman, S., Alberch, P., Camphell, J.,
Goodwin, R., Lande, R . , Raup, D. and Wolpert, L. (1Y85). Developmental
conatraints and evolution. Quart. Rev. Biol. 60, 265-287.
18 Gould, S. J . (1989). A developmental constraint in Cerion. with comments
on the definition and interpretation of constraint in evolution. Evolution 43(3),
516-539.
19 Amhros, V. (1988). Genetic basis for heterochronic variation. In: h4. L.
McKinney. ed., Heterochrony in E1dution: A Multidkciplinary Approach.
New York: Plenum Press.
20 Hunt, P., Gullisano, M., Cook, RI., Sham, W - H . , Faiella, A., Wilkinson,
U., Bonrinelli, E. and Krurnlauf, R. (1991). A distinct HOX code for the
hranchial region of the vertebrate head. Narure 353, 861-864.
Zoology, Harvard University. Cambridge, MA
02138, USA.