Full text of "Memoirs of the California Academy of Sciences"

See other formats

BIOLOGY AS HISTORY

Papers from International Conferences Sponsored by
the California Academy of Sciences in San Francisco
and the
Museo Civico di Storia Naturale in Milan

No. 2
NEW PERSPECTIVES ON THE HISTORY OF LIFE:

ESSAYS ON SYSTEMATIC BIOLOGY AS HISTORICAL NARRATIVE

(San Francisco, 21—23 June 1994)

Edited by Michael T. Ghiselin and Giovanni Pinna

nN HSONT

CiBRARIES

Memoirs of the California Academy of Sciences Number 20

New Perspectives on the History of Life:

Essays on Systematic Biology as Historical Narrative

4 EMBER'S
ENTRANCES

PARTICIPANTS
(Shown left to right in photograph)

limothy Rowe

Department of Geological Sciences and
Vertebrate Paleontology Laboratory

University of Texas at Austin

Giovanni Pinna
Museo Civico di Storia Naturale
Milan, Italy

indro Minelli
Department of Biology

University of Padova, Italy

ind Philosophy of Science Program
Cahtfornia, Davis

{ Vertebrate Zoology and
nt of Integrative Biology

Califorma, Berkeley

GRour iy
ENTRANCE =

Charlotte P. Mangum
College of Willlam and Mary

Mikhail Fedonkin
Paleontological Institute
Russia Academy of Sciences

Michael T. Ghiselin
Center for the History and Philosophy of Science
California Academy of Sciences

Robert J. O’ Hara
Cornelia Strong College
University of North Carolina at Greensboro

Nicholas D. Holland
Scripps Institution of Oceanography

BIOLOGY AS HISTORY

Papers from International Conferences Sponsored by
the California Academy of Sciences in San Francisco
and the
Museo Civico di Storia Naturale in Milan

No. 2

NEW PERSPECTIVES ON THE HISTORY OF LIFE:
ESSAYS ON SYSTEMATIC BIOLOGY AS HISTORICAL NARRATIVE

(San Francisco, 21—23 June 1994)

Edited by Michael T. Ghiselin and Giovanni Pinna

Memoirs of the California Academy of Sciences Number 20

October 4, 1996

SCIENTIFIC PUBLICATIONS COMMITTEE:

Alan E. Leviton, Editor

Katie Martin, Managing Editor
Thomas F. Daniel

Michael T. Ghiselin

Robert C. Drewes

Wojciech J. Pulawski

Adam Schiff

Gary C. Williams

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic
or mechanical, including photocopying, recording, or any information storage or retrieval system, without permission
in writing from the publisher.

Library of Congress Catalog Card Number: 96-8544]
ISBN 0-940228-43-2

TABLE OF CONTENTS
RIB BUA GB uses aya fh sagtaynns ono ataceueepcvgiy Wes ie acd ene craton ewayeveuuaiee Panes ctavetn 548 side aoa 4s OM es ee tew) do une Vil

INTRODUGIION Michael Te Ghiselintvas. samccr esac seca teint eiclcishsis = = ciel aiace eave aioe use erae |

CHAPTER 1. Robert J. O'Hara
The Space of Time: Representing the Past in the Historical Sciences. ... 2.0.0... ccc eee c cence cece nese ene 7

CHAPTER 2. James R. Griesemer
Periodization and Models in Evolutionary History ............ 0 cee cee cece cee ee tee eee ete een eect ee nee 19

CHAPTER 3. Mikhail Fedonkin
The: Oldest ‘Fossil Animals in Ecological Perspective .22......cac¢ no.ehb.s eeeee sets tees tees see ecneaeen sce 31

CHAPTER 4. Giovanni Pinna
Biogeographic Causes of Discontinuity in the Fossil Record of
the: cATMIMOMITES sa ciuia Sg dinns cad Slaw abies hie Aan iui d Sa eae argrg Rodnud, Suelo MGi ale WW and Genter dss Siaaeienayaud wich atta ae Ear bre 47

CHAPTER 5. Alessandro Minelli
Segments, Body Regions, and the Control of Development through

CHAPTER 6. Nicholas D. Holland
Homology, Homeobox Genes, and the Early Evolution of the
AV EXLED GALES es wy fectan Sasi oh tes tena cen ees cc eee odio. eouctauciees onohe ce ektvee oe vailegeiseie sep cuevaisia S66 dllataieayas 0) ss. el sb eyoues awa) au etapa oye Mia fersagns 63

CHAPTER 7. Timothy Rowe
Heterochrony of the Central Nervous System and its Effect on
IMIOKDHOP EMESIS rege creyeeyetcree nce ne raeruepe tre eas nes ae gen Wc Ne ish gorew ev ome reianey eoyeteue eens te Gus esis <eatenet et eens etree 71

CHAPTER 8. David M. Wake
Evolutionary Developmental Biology — Prospects for an
Evolutionary Synthesis at the Developmental levels... = ite <2. ecko. ena eee hnetebis eee encase ase oT,

an dnl

PREFACE

This volume is based upon a meeting held in San Francisco
on June 21 to 23, 1994 in conjunction with the Annual Con-
vention of the Pacific Division of the American Society for
the Advancement of Science. It was the second meeting deal-
ing with the general topic of biology as history to be jointly
co-sponsored by the Museo Civico di Storia Naturale in Mi-
lan, Italy and the California Academy of Sciences in San
Francisco. The first meeting was a workshop entitled “Sys-
tematic Biology as an Historical Science” held at the Milan
museum in June, 1993. The proceedings of the Milan meet-
ing, entitled “Biology as History: Systematic Biology as an
Historical Science,” edited by Giovanni Pinna and Michael
Ghiselin, have been published as Volume | of the Memorié
della Societa Italiana di Scienze Naturali del Museo Civico
di Storia Naturale di Milano. Copies of the first volume may
be purchased from the Societa Italiana di Scienze Naturali,
corso Venezia 55, 20121 Milan, Italy.

The California Academy of Sciences and the Museo Civico
di Storia Naturale are both major natural history museums,
and both are located in public parks in major cultural centers.
Both museums had to be rebuilt virtually from scratch after
having been destroyed, the one in San Francisco by the great
earthquake and fire of 1906, the one in Milan as the result

Vil

of an air raid during the Second World War. Institutions like
these benefit greatly from the exchange of ideas, as well as
specimens, and from operating on a world-wide, as well as
a regional basis. A series of international meetings that ex-
plored the prospects for doing new kinds of research in the
context of the natural history museum seemed an appropriate
way to encourage further cooperation. There has been just
enough overlap in theme and participants to allow for some
continuity. A third conference, on “The Culture of Natural
History,” has been scheduled for November, 1996, in Milan.

The Planning Committee for this meeting consisted of Mar-
valee Wake, Giovanni Pinna, and Michael Ghiselin. Among
Academy staff we should like to thank Patricia Dal Porto
for her organizational performance, and Katie Martin for her
work on getting the proceedings published. Alan Leviton
helped in ways too numerous to enumerate, but we should
at least mention the enormous amount of work he did in
setting up the meeting of the AAAS Pacific Division and
especially our part of that meeting. The Division co-spon-
sored the meeting, made its resources available, provided
valuable logistical support, and helped to attract an enthusi-
astic audience. A benefactor of the Academy, Gordon Getty,
provided funds for the conference.

Michael T. Ghiselin

INTRODUCTION

Michael T. Ghiselin

Center for the History and Philosophy of Science
California Academy of Sciences
San Francisco, California 94118

The essays presented here are all concerned with the gen-
eral topic of how systematic biology may be treated as an
historical science — one that aims, at least in the long run,
at telling the story of life on earth. Systematics is not the
only science that presents its findings in such narrative form.
Cosmology, with its tales about the “big bang” and the for-
mation of the stars and planets, aims at tracing the history
of the entire universe. Geology does likewise on a much
more local scale, and the emergence of plate tectonics in the
present century 1s a superb example of the kind of synthesis
that historical sciences can attain. Paleobiology is flourishing
these days largely because it has linked up its efforts so
effectively to the history of our planet in general.

Paleontology deals with the history of life through time,
biogeography with the same basic phenomena as reflected
in distribution of organisms in space. The various taxonomic
specialties, such as zoology and botany, although they origi-
nated without obvious reference to the history of life, were
set upon an historical foundation — “descent with modifica-
tion” —by Charles Darwin (1859). Emst Haeckel (1866)
coined the term “phylogeny” for the history of the lineage,
in contradistinction to “ongogeny” for the history of the in-
dividual organism. He evidently meant phylogeny to be an
historical science, complete with narrative accounts of what
had happened and why, and not just a bunch of tree-like
genealogical diagrams.

The present century has witnessed an increasing legitima-
tion of at least some aspects of phylogenetics. De Querioz
(1988) goes so far as to insist that descent should become
part of the “axiomatic” structure of systematic biology in
general. Nonetheless many systematists seek to decouple the
study of systematics from that of events and processes —
something that would seem very odd to a student of plate
tectonics but has strong roots in comparative anatomical tra-
dition.

Anthropology may reasonably be considered a minor
branch of systematic zoology, one that deals with the history
of our own species and that of our close relatives. Disciplines
that deal with a wide range of cultural phenomena can rea-
sonably be treated in the same spirit. Comparative linguistics
developed a phylogenetic approach before systematic zool-
ogy did, and the detailed analogies were worked out in con-
siderable detail by August Schleicher (translation 1869),
Haeckel’s linguist colleague at the university of Jena
(O’ Hara, 1992, 1993, and this volume). Lately the similarities
between cladistic methodologies and those used in disciplines
outside of biology proper have received a great deal of at-
tention (Honigswald and Wiener, 1987). Very little, however,

has been said in this literature about the narrative aspect of
systematics, and much of that for the sake of derision.

And much that gets written is grossly distorted. Cameron
(1987:238) for example, writes: “For the zoologist the re-
constructed ancestor is a fiction. It is a convenient way of
representing the organized information about the relation-
ships of the real animals in question. Nobody tries to recon-
struct a living, breathing thecodont or the protodipteron. But
the reconstructed text of the textual critic is the real article,
in a literal sense the text that Euripides wrote.” Whoever
may have provided Cameron with this misinformation, he
or she was definitely not speaking for all of us.

Of course it is not just the practitioners of historical sci-
ences who tell stories. So do poets and novelists. With sci-
entists, it really matters whether or not the stories in question
are true, and that seems to be about as good a criterion of
demarcation between science and non-science as any. To be
sure, a good poet doesn’t trade the truth for a rhyme, and a
good novel is supposed to be “true to life.” There is indeed
a certain ambiguity, in the case of the historical novel, and
in that of science fiction, wherein certain constraints of plau-
sibility are accepted as part of the genre — and if one wants
to put down serious history one of course makes the com-
parison with imaginative literature.

Historians have generated a great deal of literature attempt-
ing to show that their own enterprise is somehow different
from that of the natural sciences. We needn’t go into these
efforts in detail (for critique see Danto, 1985; Zanzi, 1991;
Laudan, 1992), but it is worth noticing that many supposed
peculiarities of history are by no means unfamiliar in what
biologists would call straight-forward historical sciences. For
example, the intentions of human beings are considered 1m-
portant causal influences. True, but the mating propensities
and other behavioral dispositions of animals in general are
an important aspect of ethology, which, after all, is largely
a comparative anatomy of inherited motor patterns.

A philosophically more pretentious claim depends upon a
criterion that purportedly demarcates that which is science
from that which is not. According to this view, science deals
only with universals, or classes; and since history deals with
individuals, or in other words with particular things, it 1s not
science. Such notions have been widely enough expressed
that it helps to examine them in a bit more detail. We should
mention at the outset that the “new ontology” that has had
so much effect on the philosophy of systematics over the
last quarter of a century arose partly in response to such
allegations. It takes the position that an individual is any
concrete, spatio-temporally restricted entity, of which an

organism is only one of a vast range of legitimate examples.
The parts of organisms — all of them, down to the ultimate
particles of matter—are individuals in this ontological
sense. So too are the earth, the solar system, the milky way
galaxy, and the entire universe. Species are of course indi-
viduals, and so too are such lineages as clades. Nor need
individuals in this sense be restricted to material objects or
groups of them: events such as the Permo-Triassic extinction
are perfectly good individuals from an ontological point of
view. Indeed, anything that may reasonably be said to have
a history is probably an individual.

The laws of nature are spatio-temporally unrestricted gen-
eralization about classes of individuals. They make no ref-
erence to any particular thing, be it to you or me or to the
moon. Rather, laws of nature are “about” classes of indi-
viduals. Advocates of the new ontology not only freely admit
that there are no laws for Homo sapiens, they use this as
evidence that species are individuals rather than classes (Hull,
1976). This turned the argument of Smart (1963) that biology
is not a science on its head, for Smart had claimed that bi-
ology has no laws, and therefore 1s not a science. In fact, it
has plenty of laws, but, like the laws of physics, these are
laws about classes of individuals (Hull, 1975; Ghiselin, 1988,
1989; Ereshefsky, 1992), not about any taxon in particular.
The obvious next step Is to insist that either the criterion in
question be accepted consistently and across the board, or
else that something better be put in its place. For the criterion
excludes not just systematic biology (including anthropol-
ogy) from the sciences, but also geology and astronomy.
Smart was not unaware of this problem and tried to wiggle
out by noting that physicists observe celestial bodies in order
to test their hypotheses about the laws of physics. It never
seems to have occurred to him that biologists observe or-
ganisms with the same end in mind.

The criterion in question does not stand up under critical
examination. Furthermore, once we have recognized the
depth of the metaphysical cleavage between individuals and
classes, we are in a much better position to appreciate what
it is that the historical sciences are all about. Without them,
science 1s a purely abstract matter that deals with everything
in general and nothing in particular. The traditional distine-
tion, which goes back to Wilhelm Windelband, between
idiographic (etymologically “person writing’) and no-
mothetic (etymologically “law propounding”) sciences has
long recognized the distinction. And yet the habit of treating
history and law as if they were the domains of strictly sepa-
rate disciplines is to make an ontological distinction function
as a spurious guideline for scientific practice. Studying the
laws and the particulars together as part of a single research
program has proven such a remarkably successful stratagem
in the history of the sciences that one would think it hardly
needs justification. And yet there are many who insist that
the two ought to be kept more than just conceptually distinct.
They present us with the sort of “naive inductionist” model
that philosophers of science long ago ceased to take seriously.
The idea is that one starts with description of brute facts,
puts the facts together into ever broader generalizations, and
finally arrives at the laws — or perhaps the narratives. The

INTRODUCTION

possibility that one might proceed by reiterated testing of
hypotheses is simply left out of consideration. Given such
an epistemology, it 1s easy to see why such narrative is so
rarely considered as a constitutive part of the process of dis-
covery and testing. But if possible narratives or scenarios
and the laws of nature are introduced from the outset, we
get a very different picture of the dynamics of investigation.
This is not the appropriate place to do more than suggest
the role that laws of nature and other generalizations play
in the evaluation of reconstructions in the historical sciences.
We should emphasize, nonetheless, that methodologists are
well aware of them. In just about any science, as in everyday
life, we know that certain classes of events are more probable
than others, and that some are downright impossible. Any
proposition that is necessarily true irrespective of time and
place must be true of any particular event. So if a scenario
logically entails the falsehood of such a proposition we con-
sider it good evidence for rejecting that scenario. Organisms
simply must have a source of energy and they cannot function
as perpetual motion machines, so the larvae that subsist en-
tirely upon stored yolk are not recapitulating a stage in which
their ancestors did not feed at all! By the same token, we
treat scenarios as more plausible, in proportion as they invoke
fewer events that the laws imply are improbable. Of course
we might opt to treat such contradiction as a refutation of
the law of nature in question, but reasonable scientists gen-
erally ask for further evidence before doing anything so rash
if the laws are well supported by empirical evidence. Fur-
thermore the research program can be designed to test both
the laws and the scenarios. (Yes, there is indeed a danger
of circularity here — no, it is not an insuperable difficulty.)

In earlier works | have explained how this is accomplished
in geology and in linguistics, as well as comparative anatomy
(Ghiselin, 1969, 1972), presenting some straight-forward ex-
amples. Stratigraphic geology establishes the sequence in
which rocks are deposited largely by means of the principle
of superposition: the oldest rocks are on the bottom, the
youngest ones on top. This is by no means an a priori as-
sumption in any proper, philosophical sense, because It 1s
justified by experience and erroneous applications of it can
be corrected by empirical data. There is good reason why
the water that deposits sediments floats above the pre-exist-
ing strata. Furthermore, if the strata should happen to be
overturned, one can generally tell that this has happened be-
cause cracks in dried mud, wave patterns, varves, and other
features necessarily have a definite orientation with respect
to the force of gravity (Shrock, 1948). The laws governing
language are a bit less familiar, but to give one example,
Wang (1987:248) remarks that “it 1s more likely for two
languages independently to undergo the same natural change
than for them independently to undergo the same unnatural
change. Consequently, the sharing of unnatural changes 1s
more diagnostic of shared history.”

Narratives, unlike mere chronicles, do not simply describe
what happened, they attempt to explain it. Scientific expla-
nation is by no means an uncontroversial topic among phi-
losophers, and perhaps the less we say about their difficulties
with it the better (see the anthology edited by Ruben, 1993).

MICHAEL T. GHISELIN

But we should at least remark that, according to the “covering
law model” which is largely associated with the names of
Hempel (1942) and of Popper (1962), historical explanation
somehow attempts to link up the particular events with at
least one law of nature. This model seems — I repeat seems
— unsatisfactory on a number of accounts, not the least of
which is that the laws which are invoked as predicting what
goes on seem much more convincing when applied retrospec-
tively.

Be this as it may, Darwin, after completing a massive trea-
tise on the systematics of barnacles, went on to present what
he considered compelling evidence for evolution in general
and for natural selection as its basic mechanism in particular,
structuring his argument on the basis of its explanatory
power. Among his contemporaries, some, including Herbert
Spencer, agreed with Darwin that a theory that explains such
a wealth and diversity of facts must contain a large measure
of truth; but many others, such as Thomas Henry Huxley,
did not like that manner of reasoning. Opinion remains di-
vided. However, it does seem that many, perhaps the vast
majority, of evolutionary biologists, and practitioners of other
historical sciences as well, would agree with the famous
aphorism of Dobzhansky (1973) that “Nothing in biology
makes sense except in the light of evolution.”

So we have very good reason to take the narrative aspect
of evolutionary biology very seriously, to assess its possible
strengths and limitations, and to see what might be done
with it in future research. But as Landau (1991) points out,
in a very interesting study on the history of narratives of
human evolution, scientists themselves have said very little
about the narrative aspects of their own research.

The present volume, although it 1s not exactly what Landau
had in mind, may help to supply the deficiency. The studies
presented here provide some conceptual background at the
beginning, then go on to discuss some particular cases,
mainly having to do with the fossil record and the embryo-
logical aspects of comparative anatomy. The chapters turn
out to fit together into a coherent whole to a far greater
extent than the organizers or contributors originally intended.

Historical narratives are by no means random collections
of events, but focus upon details that are deemed to have
particular causal significance. O’Hara makes this abundantly
clear in his essay on how the past is represented in the his-
torical sciences. Maps provide a particularly good example
of how matters get generalized, simplified, and perhaps dis-
torted. He gives some amusing examples of maps that were
intended as jokes, ridiculing the provincial attitudes of those
who were supposed to draw them. Perhaps equally ridiculous,
but something we ought to take seriously, is the manner in
which systematic biology is all too often depicted in text
books, as if our own lineage were the only interesting object
of study.

Griesemer addresses a somewhat related issue in his dis-
cussion on periodization. It seems clear, if not altogether
obvious, that a choice will have to be made as to how a
temporal sequence is to be subdivided if the past is to be
treated as a classification, and not just a list, of events. The
obvious place to put the breaks are the events that supposedly

have important causal significance. Branching points in ge-
nealogies are an obvious example, and those systematists
who have wanted to represent more than just branching
points in their classifications have traditionally used cate-
gorical rank to indicate major changes in structure and func-
tion. Griesemer discusses how embryology can be periodized
in various ways, some of which are able to link the historical
narrative up to morphogenetic causes, a theme that gets taken
up again later in the book.

The next two chapters deal with the fossil record. Here,
the periodization is largely in terms of external, ecological
causes. Fedonkin considers early stages in the history of mul-
ticellular animals, and develops an ecological scenario that
helps to account for their peculiar properties. We should em-
phasize that he presents data that seem inconsistent with other
possible scenarios, including those which invoke total anoxia.
Perhaps the most original aspect of his contribution is the
attempt to show how the animals have interacted with one
another and with the changing abiotic environment through
time; it is hard to imagine how it would be possible to infer
such interactions on the basis of cladistic data by themselves.
The recent advances in paleontology which he reviews un-
derscore the difficulties that have attended the reconstruction
of fossil organisms and their conditions of existence. Doing
the job properly requires a great deal of thinking, both critical
and imaginative.

Pinna’s scenario is largely a biogeographic one, and has
a strong ecological component. It provides a most instructive
example of what can be done with a good fossil record when
it is properly analyzed. A great deal of theory in paleontology
has been devised to explain what are really artifacts of pres-
ervation and misunderstandings of taxonomy. Pinna presents
an account of ammonites diversifying in a stable environment
and giving rise to descendant populations in a less stable
one, where they fail to diversify and found lineages. This
scenario accords nicely with the sort of diversity model that
Darwin and others have proposed and also with the centers
of origin concept to which vicariance biogeographers are so
often hostile. But it does not depend on that kind of evolu-
tionary model for its empirical justification. Having provided
grounds for rejecting the reality of the phenomena upon
which certain macroevolutionary theories have been based,
he goes on to confront some claims about the underlying
developmental mechanisms with hard data.

Thus we get a smooth transition to the following chapters,
in which embryology becomes the dominant explanatory
theme. Minelli considers some correlations among the vari-
ous parts of the body as arthropods have become more com-
plex, and includes the developmental stages in his analysis.
There seems to be a definite limit to the number of kinds
of parts that an organism can contain, with arthropods for
example never having more than four or five tagmata (body
regions). He suggests various reasons for this, including con-
flicting “design” constraints as with increasing numbers of
parts. Since the unity of the historical sciences 1s a major
theme in this work, it seems worth while to draw attention
to a similar feature of language. Ethnobotanists and other

students of folk taxonomy have found that the classification

systems they have studied are remarkably uniform in having
a hierarchical arrangement with only about five levels, or as
systematists call them, “categories” (Berlin, 1992). It would
seem that there is also a definite upper limit to the amount
of information that can readily be processed by the unaided
human brain. Perhaps some very general laws and principles
are applicable.

Holland finds it necessary to discuss the various concepts
of homology, in order to disassociate his approach from those
of certain other workers. The lack of consensus among com-
parative biologists these days as to what so basic a term as
“homology” is supposed to mean reflects a continued failure
of the intellectual community to come to grips with the fun-
damental distinction between classes and laws on the one
hand, and individuals and history on the other. Homology
and analogy alike are terms for the correspondence of parts
of organisms. In the case of homology, the organisms them-
selves are by definition parts of lineages, and the criterion
of identity is community of descent. In the case of analogy,
the organisms are by definition members of classes, and the
parts are equivalent because of some nonhistorical cause,
such as convergence resulting from similar function. Confu-
sion results when people try to mix the two.

Holland’s preference for the historical homology concept
makes perfectly good sense in terms of the rationale for his
research. A very elegant technique allows him to localize a
particular kind of tissue component that for good reasons 1s
believed not likely to change its relative position in the body.
There is a good precedent for invoking such a criterion, in-
sofar as comparative anatomists have long recognized that
the Innervation of organs 1s much less labile than many other
features of the body. The new results are most impressive,
and when they disagree with received opinion, they are all
the more impressive. The alternatives are very plausible, and
indeed represent views that have been held by a minority of
workers, but perhaps have not received the amount of atten-
tion they deserve.

Impressive new techniques are also applied by Rowe to
elucidate the history of the mammalian skull. But once again,
we have technology in the service of problems with a long
history. Rowe emphasizes the point that modern cladistic
techniques have helped to correct some serious misconcep-
tions about the supposed polyphyly of mammals and the par-
allelism of important mammalian characters. On the other
hand his own approach has itself tended to break with cladis-
tic tradition, especially insofar as it has emphasized func-
tional analysis. He and his collaborators (Gauthier, Kluge
and Rowe, 1988) were able to show that a large number of
features of terrestrial vertebrates are causally interdependent:
they make it possible for the animals to run and breathe at
the same time. This discovery tended to support the notion
that in phylogenetic analysis characters ought to be weighted
on the basis of whether or not they are functionally interde-
pendent (Gosliner and Ghiselin, 1984). But how to accom-
plish such weighting in practice remains a serious bone of
contention

The problem that Rowe addresses, namely, how the mam-
malian inner ear arose, is a traditional one, and the part of

INTRODUCTION

the traditional picture that remains intact is one of change
in both structure and function with continuity. Part of the
jaw gradually became incorporated into the ear. It is a fine
example of the principle of succession of functions (Funk-
tionswechsel) first propounded by Dohrn (1875; translation
in Ghiselin, 1994). Physiological work done outside of a
phylogenetic context helps to elucidate the function of the
apparatus in question, but only becomes explanatory when
incorporated into an historical narrative. Rowe enriches the
scenario by adding a morphogentic analysis to the list of
causes of change. He is even able to specify how some of
the physical forces operate in developing embryos. All sorts
of interesting and quite unexpected causal relationships
emerge from this research, including the point that the sepa-
ration of the auditory apparatus from the jaw 1s not a post-
adaptation, but instead is a byproduct of the enlargement of
the brain.

Wake takes on the challenge of prognosticating how we
might be able to come up with a unitary approach, or a
synthesis, of evolutionary and developmental biology. As he
points out, this kind of approach has become relatively fash-
ionable of late, partly because of the influence of a book by
Gould (1972). Yet I hasten to point out that it is a traditional
theme in evolutionary biology, albeit one that has been ne-
glected in elementary text books. Darwin (1868) devoted an
entire book to that very theme — though Gould (1972) does
not even cite it. Among later writers we should definitely
mention Sewertzotf (1931) and Schmalhausen (1946), who
had considerable influence upon Dobzhansky, Mayr, Rensch,
and other architects of the Synthetic Theory.

One reason why, except to some extent in Russia, such
developmental themes have been treated as marginal has
been repeated efforts to develop alternatives to natural se-
lection as the fundamental evolutionary mechanism. Hence
we get anti-evolutionists (in a broad sense) trying to treat
evolution as if it were ontogeny writ large, often under the
rubric of “orthogenesis.” The idea was to explain change in
terms of “laws” rather than “chance” — laws that might or
might not have been supernaturally ordained (Hertwig,
1922). “Chance” in this context 1s a pejorative synonym for
“historical contingency” or what most systematists these days
consider the main topic of their research. Nonetheless, as
Wake emphasizes, a lot of evolutionary biologists tend to
emphasize what he and many others call “pattern” over proc-
ess. This I would take as symptomatic of a static world view,
as is the concern for such idealistic notions as the “Bauplan”
even when translated into another jargon, as in the “zootype”
of Slack et al. (1993) to which Wake and also Holland refer.
Wake emphasizes the need to separate process from pattern
and to keep the two conceptually distinct; but one wonders
if what is being confused is not process and product. Indeed
itis far from clear to me what authors mean when they refer
to pattern and process. By recapitulation, | do not mean a
pattern, but a developmental process that has a pattern similar
to that of an evolutionary one. The term “reverse recapitu-
lation” advocated by Wake and his collaborators has received
a lot of sales-resistance for that very reason. It suggests that
the stages have somehow been turned around backward, like

MICHAEL T. GHISELIN

a motion picture being shown as the film is rewound.

Wake discusses what he calls the “persistence” of “char-
acters,” providing a new way of talking about what used to
be discussed in terms of “conservativeness” or “lability” in
evolution. Some things clearly have changed more than oth-
ers in the course of phylogenetic history. In principle, if we
understood the underlying mechanisms of development, we
would be in a much better position to estimate the lability
in question and consequently to reconstruct the history of
life. However, it would seem that more than one phenomenon
is involved here. On the one hand we have the intrinsic as-
pect, in the sense that the developmental processes them-
selves are not readily changed, as suggested by Minelli in
the present volume. On the other hand, we have an extrinsic
aspect, in the sense that certain changes have not been se-
lectively advantageous. A feature might persist unchanged
over long periods of time simply because, as a matter of
contingent fact, the organisms remained under conditions of
existence where it was advantageous.

Not only is the conservativeness in question largely an
extrinsic and contingent matter, it 1s obviously a matter of
degree, so that even if a kind of inertia results from devel-
opmental constraints, there is no reason to presuppose that
sufficiently strong selection pressures cannot overcome It.
There has been an unfortunate tendency of late to express
such considerations in pre-evolutionary, idealistic terms, hav-
ing recourse to such notions as Baupldane where a real evo-
lutionist would speak of common ancestors. It seems to me
an unfortunate source of confusion when Gould (1983) tries
to justify his version of typology by trying to lodge the
equivalent of a Platonic Idea in the germ rather than in the
Mind of God.

So, rather than talk about abstract patterns and schematic
diagrams, it might be better to focus our attention upon con-
crete beings that really existed and upon historical events
that actually took place. Such a narrative account would be-
gin with common ancestors, and would trace the changes
that have gone on in the various descendant lineages. In such
narrative embryology could play a crucial role, for as the
organisms change so must the developmental mechanisms
that give rise to them. And our understanding of those mecha-
nisms could play a crucial role in both discovery and expla-
nation. Indeed it already does, as can be seen in the more
particular studies by Minelli, Holland and Rowe, as well as
in the more general overview provided by Wake.

Looking to the future, we might consider whether to place
an even stronger emphasis upon the scenario. Instead of in-
terpreting the tree-like diagram as a representation of char-
acters, treat it as made up of historical events. In that case
the emphasis upon process becomes all the more obtrusive,
and our thinking becomes increasingly Darwinian.

Literature Cited

Berlin, B. 1992. Ethnobiological Classification: Principles of Cate-
gorization of Plants and Animals in Traditional Societies. Prince-
ton University Press, Princeton.

Cameron, H. D. 1987. The upside-down cladogram: problems in
manuscript affiliation. Pages 227-242 in H. M. Hoenigswald and

L. F. Wiener, editors. Biological Metaphor and Cladistic Classi-
fication: an Interdisciplinary Perspective. University of Pennsyl-
vania Press, Philadelphia.

Danto, A. C. 1985. Narration and Knowledge (Including the Integral
Text of Analytical Philosophy of History), 2nd Edition. Columbia
University Press, New York.

Darwin, C. 1859. On the origin of species by means of natural
selection, or the preservation of favoured races in the struggle
for life, Ist Edition. John Murray, London. (Sixth Edition 1872.)

——.. 1868. The Variation of Animals and Plants under Domes-

tication, Ist Edition. John Murray, London. (2nd ed. 1875.)

De Queiroz, K. 1988. Systematics and the Darwinian revolution.
Philosophy of Science 55:238—259.

Dohrm, A. 1875. Der Ursprung der Wirbelthiere und das Princip
des Functionswechsels. —Genealogische Skizzen von Anton
Dohrn. Wilhelm Engelmann, Leipzig.

Ereshefsky, M. 1992. The historical nature of evolutionary theory
Pages 81-99 in M. H. Nitecki and D. V. Nitecki, editors. History
and Evolution. State University of New York Press, Albany.

Gauthier, J., A. G. Kluge, and T. Rowe. 1988. Amniote phylogeny
and the importance of fossils. Cladistics 4:105—209.

Ghiselin, M. T. 1969. The Triumph of the Darwinian Method, Ist
Edition. University of California Press, Berkeley.

1972. Models in phylogeny. Pages 130-145 in T.J.M.
Schopf, editor. Models in Paleobiology. Freeman and Cooper,
San Francisco.

—. 1988. The individuality thesis, essences, and laws of nature.
Biology and Philosophy 3:467-474.

. 1989. Individuality, history and laws of nature in biology.
Pages 53-66. in M. Ruse, editor. What the philosophy of biology
is. Kluwer Academic Publishers, Dordrecht.

—. 1994. The Origin of Vertebrates and the Principle of Suc-
cession of Functions. Genealogical Sketches by Anton Dohrn.
1875. An English translation from the German, introduction and
bibliography. History and Philosophy of Life Sciences 16:5—98.

Gosliner, T. M., and M. T. Ghiselin. 1984. Parallel evolution in
opisthobranch gastropods and its implications for phylogenetic
methodology. Systematic Zoology 33:255—274.

Gould, S.J. 1977. Ontogeny and Phylogeny. Harvard University
Press, Cambridge.

. 1983. Irrelevance, submission, and partnership: the chang-
ing role of palaeontology in Darwin’s three centennials, and a
modest proposal for macroevolution. Pages 347-366 in D.S.
Bendall, editor. Evolution from Molecules to Men. Cambridge
University Press, Cambridge.

Haeckel, E. 1866. Generelle Morphologie der Organismen. Allge-
meine Grundziige der organischen Formen-Wissenschaft, mecan-

isch begriindet durch die von Charles Darwin reformirte Descend-
enz-Theorie. Verlag von Georg Reimer, Berlin. (Taf. I-IIl, I-VI)

Hempel, C. G. 1942. The function of general laws in history. The
Journal of Philosophy 39:35-48.

Hertwig, O. 1922. Das Werden der Organismen: zur Widerlegung
von Darwins Zufallstheorie durch das Gesetz in der Entwicklung,
3rd Edition. Verlag von Gustav Fischer, Jena.

Hoenigswald, H.M., and L. F. Wiener, editors. 1987. Biological
Metaphor and Cladistic Classification: an Interdisciplinary Per-
spective. University of Pennsylvania Press, Philadelphia.

Hull, D. L. 1975. Central subjects and historical naratives. History
and Theory 14:253—274.

_ 1976. Are species really individuals? Systematic Zoology
25:174-191.

Landau, M. 1991. Narratives of Human Evolution. Yale University
Press, New Haven.

Laudan, R. 1992. What’s so special about the past? Pages 55-67
in M.H. Nitecki and D. V. Nitecki, editors. History and Evolu-
tion. State University of New York Press, Albany.

O’Hara, R. J. 1992. Telling the tree: narrative representation and
the study of evolutionary history. Biology and Philosophy 7:135—
160.

1993. Systematic generalization, historical fate, and the
species problem. Systematic Biology 42:231—246.

INTRODUCTION

Ruben, David-H., editor. 1993. Explanation. Oxford University
Press, Oxford.

Schleicher, A. 1869. Darwinism Tested by the Science of Language.
John Camden Hotten, London, Transl. Alex. V. W. Bikkers.

Schmalhausen, I. I. 1946. Faktori Evolutsii (Teortya Cta-biliziryut-
schego Otbora). Izdatelstvo Akademi Nayk, Moskva.

Sewertzoff, A. N. 1931. Morphologische Gesetzmiassigkeiten der
Evolution. Verlag von Gustav Fischer, Jena.

Shrock, R. R. 1948, Sequence in Layered Rocks: a Study of Features
and Structures Useful for Determining Top and Bottom or Order
of Succession in Bedded and Tabular Rock Bodies. McGraw-Hill,
New York.

Slack, J.M. W., P. W. H. Holland, and C. F. Graham. 1993. The
zootype and the phylotypic stage. Nature 361:490-492.

Smart, J. J.C. 1963. Philosophy and Scientific Realism. Routledge
and Kegan Paul, London.

Wang, W. S.-Y. 1987. Representing language relationsips. Pages
243-256 in H. M. Hoenigswald and L. F. Wiener, editors. Bio-
logical Metaphor and Cladistie Classification: an Interdisciplinary
Perspective. University of Pennsylvania Press, Philadelphia.

Zanzi, L. 1991. Dalla Storia all’Epistemologia: lo Storicismo Scien-
tifico. Principi di una Teoria della Storicizzazione. Jaca Book,
Milano.

MAPPING THE SPACE OF TIME:
TEMPORAL REPRESENTATION IN THE HISTORICAL SCIENCES

Robert J. O’Hara

Cornelia Strong College
100 Foust Building
University of North Carolina at Greensboro
Greensboro, North Carolina 27412

As to the propriety and justness of representing sums
of money, and time, by parts of space, tho’ very readily
agreed to by most men, yet a few seem to apprehend
that there may possibly be some deception in it, of
which they are not aware... .

William Playfair, 1786, in Tufte (1983:52)

Introduction: The Palaetiological Sciences

William Whewell (1794-1866), polymathic Victorian sci-
entist, philosopher, historian, and educator, was one of the
great neologists of the nineteenth century. Although Whew-
ell’s name is little remembered today except by professional
historians and philosophers of science, researchers in many
scientific fields work each day in a world that Whewell
named. “Miocene” and “Pliocene,” “uniformitarian” and
“catastrophist,” “anode” and “cathode,” even the word “scl-
entist” itself — all of these were Whewell coinages. Whewell
is particularly important to students of the historical sciences
for another word he coined, one that was unfortunately not
as successful as many of his others because it is difficult to
pronounce. This word, “palaetiology,” was the name Whew-
ell gave to the class of sciences that are concerned with
historical causation: the class we might today refer to as
historical sciences. Although the disciplines Whewell in-
cluded under the heading of palaetiology might seem to cut
across the conventional academic boundaries of his day and
ours — his exemplars were geology and comparative philol-
ogy —all these fields may nevertheless be examined to-
gether, Whewell argued, because of their common interest
in reconstructing the past. Just as we may look back, he said,

towards the first condition of our planet, we may in
like manner turn our thoughts towards the first condi-
tion of the solar system, and try whether we can discern
any traces of an order of things antecedent to that
which is now established; and if we find, as some
great mathematicians have conceived, indications of
an earlier state in which the planets were not yet gath-
ered into their present forms, we have, in pursuit of
this train of research, a palaetiological portion of As-
tronomy. Again, as we may inquire how languages,

New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

and how man, have been diffused over the earth’s sur-
face from place to place, we may make the like inquiry
with regard to the races of plants and animals, founding
our inferences upon the existing geographical distri-
bution of the animal and vegetable kingdoms: and thus
the Geography of Plants and of Animals also becomes
a portion of Palaetiology. Again, as we can in some
measure trace the progress of Arts from nation to na-
tion and from age to age, we can also pursue a similar
investigation with respect to the progress of Mythol-
ogy, of Poetry, of Government, of Law... . It is not
an arbitrary and useless proceeding to construct such
a Class of sciences. For wide and various as their sub-
jects are, it will be found that they have all certain
principles, maxims, and rules of procedure in common;
and thus may reflect light upon each other by being
treated together.

(Whewell, 1847, 1:639-640)

This paper is an essay on the palaetiological sciences, dedi-
cated to Whewell on the bicentennial of his birth, an essay
that examines some of the “principles, maxims, and rules of
procedure” that these sciences have all in common. Its first
purpose is to demonstrate the continuing validity of Whew-
ell’s classification of these sciences through a study of his-
torical representation in three different palaetiological fields:
systematics, historical linguistics, and textual transmission.
Its second purpose is to continue the development of an ex-
tended analogy between historical representation and carto-
graphic representation that | began in an earlier paper
(O’Hara, 1993), an analogy that makes especially clear the
common representational practices that are found throughout
palaetiology.

To set the stage for what is to follow, I offer here three
diagrams, one each from the different palaetiological fields
of systematics, historical linguistics, and textual transmission
or stemmatics, three diagrams all drawn independently within
40 years of each other in the mid-nineteenth century. The
first of these (Fig. 1), familiar to all evolutionary biologists,
is Darwin’s tree of descent from the Origin of Species (1859),
The vertical axis of this diagram represents time, while each
horizontal line marks an interval of some number of genera-
tions: a thousand, or a million, or a hundred million
(1859:116—126). Figure 2 is less familiar, even to specialists
in the field from which it comes, historical linguistics. This

1996
California Academy of Sciences

ROBERT J. OHARA

“suolesiauas
JO JOQUINU ARQIqde SLOS JO} ZuIpurys UOISIAIP YOva UIA ‘at sjuUasaidar srxv PEITUIA UL “(GSR 1) Sa2dg fo WBC s UIMIeG WI aft] JO dda) feoHayjodAy ayy “| AWNOL|

Tapreymary TT SoM M
\ H H i H ‘ x
é ' ‘ ' i ’ rt ‘ \
, 1 ‘ ' ' Hy ‘ N
/ i i ‘ ' , ' . 4
, ’ ‘ ’ \ ‘ \
, ' 4 4 ‘ ‘ : \ \
/ ‘ \ ' \ Fy Hy \ .
/ Y \ 1 f H \ %
\ 1 n
/ ' ' / H \
/ \ ' .
' ' / i ‘ ‘
‘ ' ‘ J i} ' ' ’ \
/ \ ' ‘ ‘ ‘ .
d ' ‘ ' ‘ / ‘ N
F, , \ ' ' ' ' \ \
/ ' \ ' ' ' ' ‘ \
/ ' ‘ ' ‘ / ' \ .
/ ! ‘ ' H , ' \ .

=
= e . + =
I ee 77 jo
on - —— “ + . = - -
22 =) 7 z 1D
= r. = + =
piel we 2 7] pt 7B
er > 7 7. =
AI ae: yu a? 2P 4?
a cs
aX ee BD yl vf oe oP
= =
TA 9x g7 gt oF of go?
z 7 4
TIA 5 wm a7 gut Py! Pat zd gn
ITLA 7 ET PY) 7 n uw BE Tg?
es 8 ’ ° s/ A oJ °
XI en o” 7] Py) 6?
x ar* ye: aA orl of ut or?
SSS SS ae = + a - = = et en oe
Por H or fy foie 4d
Lea ee ' rot \ / ‘ oy ‘
es oe U ies \ , ‘ yt ‘\
i; ' \ : ' , ' \ D ‘ r] ' \
‘ m4 ! ' \ ! ' \ / Hi / LU ‘
ae Se ee - a a ead ae a - j——+ - -p- at
? : 3 ! i bf ‘ ri ' \ : ‘ ¥ ' ‘
HY \ ’ 1 \ ‘ t ' \ ' 4 ri H \
i ' \ fi ' ‘ i / H \ / \ , H \
‘ i \ ' ' ‘ H . H nN ' \ / | ‘
1) Caer eee + - : {——} —-— — - + + + + 7 + —; t= =
’ ‘ ' | \ ' . , '
, / . D ' ‘ 4
/ 1 \ i ' \ ' ' ' \ / \ 4 ' \
' ' ‘ ’ ' , ‘ ’ * | ‘ ry ’ ' ‘
/ ' \ hs ' ‘ ' / ' \ / 1 i H ‘
| eee era Fenn Sees —_ a 4 : . 7 ;
AX "y= wf wt wi ar wl Pd er? er? orS +79 ae #7 ?

TEMPORAL REPRESENTATION

gor

kraj.

diagram is one of the first trees of language “phylogeny”
and it was drawn, like Darwin’s tree in the 1850s, by the
Bohemian historical linguist Frantisek Celakovsky (Cela-
kovsky, 1853; Priestly, 1975). Figure 3 is a third “tree of
history”: the first stemma of manuscript transmission, pub-
lished by Carl Johan Schlyter in 1827 (Collin and Schlyter,
1827; Holm, 1972). As in Darwin’s diagram, the vertical
axis represents time, and each horizontal line stands for a
specific time interval (25 years in this case). What these
three diagrams illustrate is that three palaetiological sciences
——~ systematics, historical linguistics, and textual transmission
— though they function independently, all produce results
of the very same sort using many of the same procedures
of inference: they all produce trees of history showing
branching sequences of ancestry and descent.

In each of these fields a great deal of attention has been
given to the methods of historical reconstruction, particularly
so in recent years in systematics, where attention has also
been given to the historical character of the discipline as a
whole (O’Hara, 1988a; de Queiroz, 1988; Ghiselin, 1991).
But in contrast to the amount of attention that has been given
to historical reconstruction (e.g. Sober, 1988), very little has
been written in any of these fields about the problems of
historical representation. Given that we have knowledge
about events that took place in the past— the geological
past, or the evolutionary past, or the linguistic or textual past
—how do we represent, how do we communicate that
knowledge? In particular, how do we use diagrams, which
are two-dimensional, spatial representations, to depict the
temporal relationships of events in time?

FIGURE 2. A genealogy of the Slavic languages drawn by Frantisek Celakovsky at Prague about 1852 and published in 1853 shortly after
g g) guag b : g I j

his death (Celakovsky, 1853; Priestly, 1975). For the only language tree earlier than Celakovsky’s (a diagram drawn around 1800 by Félix
Gallet) see Auroux (1990).

It might seem that historical representation (as opposed to
historical reconstruction) 1s unproblematic: historical scien-
tists just draw diagrams that illustrate what they know. His-
torical representation is a more subtle activity than one might
suspect, however, and I want to demonstrate this by com-
paring historical representation — the representation of
events in time — with cartographic representation — the rep-
resentation of objects in space, as we see in ordinary geo-
graphical maps. Maps might also seem to be completely un-
problematic representations of the world, but in fact they too
are rather more subtle than one might expect. In making this
comparison | will draw heavily on the work done by carto-
graphic theorists (Toulmin, 1953, Robinson and Petechenik,
1976; Gould and White, 1986; Buttenfield and McMaster,
1991; Monmonier, 1991; McMaster and Shea, 1992), as well
as some of my own earlier work on diagrams in systematics
(O’Hara, 1988b, 1991, 1992, 1993).

Maps as Spatial Representations

Maps are representations of objects in space, and they suc-
ceed as representational devices because they are selective:
because they omit a great deal of information that map-mak-
ers in fact have. Some imaginary Ideal Map that included
literally everything in the territory it represented would be
useless, because the territory itself could serve just as well
(Crampton, 1990; O?Hara, 1993). Cartographers call the
process whereby the world is reduced to a map, or a complex
map reduced to a simpler map, cartographic generalization.
The most basic element of the generalization process is the

simple deletion of certain objects from the map, objects that
exist on the earth but that will not appear on the map. But
many other processes are involved in generalization as well,
beyond the simple deletion of objects. For example, areal
features may have their outlines simplified, and linear fea-
tures may be smoothed or enhanced (Fig. 4; Monmonier,
1991). A surprising element of generalization 1s “feature dis-
placement”: when two objects are so close together on a

en)

Nchema Cognattonts

l
l
l
|
|
| Codtcum manuse.
]
|
]
|
l

1700.
ee 1
-4

!
|
|
[
|
|
|
|
|
|
|
|
|
|
{
|
|

N 3166. Arce. _ 4

hGURLE 3. Carl Johan Sehlyter’s stemma of a group of medieval
Swedish legal texts, the first diagram of textual transmission ever
published (Collin and Schlyter, 1827; Holm, 1972). Notice the re-
markable similarity to Darwin’s evolutionary diagram, with the ver-
tical axis representing absolute me and horizontal lines indicating
time intervals

ROBERT J. OHARA

map that they are difficult to distinguish, and yet both must
be included, the two objects may be nudged apart slightly
(Fig. 4). This has the effect of warping the scale of the map
in the vicinity of the displaced objects: a unit of distance on
the map in that region corresponds to a shorter distance on
the ground than does the same unit of distance on another
part of the map.

Cartographic generalization is a concept that has been de-
veloped and applied in the context of geographic maps, and
one might think at first that such a concept would have little
relevance to representations of history — to representations
of events in time rather than objects in space. But upon re-
flection it is evident that we often speak of space and time
in the same terms, and so ideas that apply in one domain
might well be useful in the other. “Short” and “long” are
adjectives that apply to “lengths” of both space and time.
We speak of “deep” time and the “distant” past. And in
answer to the question “How far is it to the city?” one Is as
likely to be told “two hours” as “100 miles.” In view of
the similarity between the language of space and the language
of time, then, let us see if the notion of generalization can
be applied with as much success to representations of events
in time (events as they are reconstructed by palaetiologists)
as it has been by cartographers to representations of objects
in space.

The Space of Time

Let us begin with the simple case of an historical diagram
that is strikingly cartographic in character (Fig. 5). This dia-
gram of phylogeny from Hennig’s well-known systematics
text (1966) shows a sequence of events at three different
temporal scales, two of them by means of insets, just as a
city map might have an inset to show the city center and
another to show the surrounding region. Apparent in this
diagram is a temporal equivalent of what cartographers call
aggregation, as when several small lakes are shown on a
map as one larger lake. Temporal aggregation ts manifest
here in the representation of several generations of individual
organisms by a single circle in the inset on the right. Simi-
larly, even though each individual organism itself has a tem-
poral dimension (its life span), each is reduced in Figure 5
to a single symbol without temporal extent. This latter phe-
nomenon is called symbolization in cartography, when an
object that occupies a definite geographical area (a city, for
example) is reduced to a single symbol such as a dot.

In the broad spirit of palaetiology, it is important to realize
that these phenomena of temporal generalization are not re-
stricted to evolutionary trees alone, as can be seen in Figures
6 and 7, two recently-published diagrams of the history of
the Germanic languages. Figure 6 shows a simplified (highly
generalized) version of the entire Germanic tree, ending in
the three branches of East Germanic, North Germanic, and
West Germanic, the last of these being the branch that in-
cludes modern English. Figure 7 shows an enlargement of
the West Germanic branch alone: the single lower right
branch of Figure 6 corresponds to the entirety of Figure 7,

just as an irregular polygon representing the city of San Fran-

TEMPORAL REPRESENTATION

Simplification

Selection

Displacement

FIGURE 4. Some elements of cartographic generalization, redrawn after Monmonier (1991)

tokogenetic
relationships

ontogenetic
relationships

species

phylogenetic
relationships

species

individual

FiGure 5. A hypothetical phylogeny, after Hennig (1966). Three different degrees of generalization are shown: the central portion of the
diagram is resolved to the level of individual organisms, while the inset at the bottom shows the life stages of one individual, and the inset at
the right shows a coarser view of the species as a whole. See O'Hara (1993) for further discussion of diagrams of this type, and see Maddison

and Maddison (1992:26) for an additional example

cisco on a map of California would correspond to an entire
San Francisco street map.

As a representation of objects in space, any geographical
map can be generalized in a number of different ways. We
could take a detailed map of San Francisco and generalize
it into a map showing only the subway lines, or only the
railroads, or only the public streets and nothing else. Simi-
larly, any given representation of events in time, such as an

evolutionary tree, can also be generalized in a number of

different ways (O’Hara, 1993). And just as different gener-

alizations of a map may give the viewer different senses of
a particular territory — one that showed all the parks might
give a different impression from one that showed only high-
ways and railroad tracks (Monmonier, 1991) — so also dif-
ferent generalizations of a detailed sequence of events in
time may give the viewer different senses of what took place
within a particular temporal space. Different generalizations
of the history of life, for example, may give the impression
that evolution is either directed or diversifying (O’Hara,
1992, 1993).

2 ROBERT J. OHARA

Proto-Gmc

EGmec NWGmc

NGmc WGmc

FIGURE 6. A simplified (highly generalized) history of the Germanic languages, after Barber (1993). The ancestral Proto-Germanic language
is shown dividing into East and Northwest Germanic branches, the latter dividing again into North and West Germanic. The West Germanic
branch, of which modern English is a part, is shown in greater detail in Figure 7

* Proto-WGmc

100. = ‘Ingvaeonic’ ‘Istvaeonic’ ‘Erminonic’

200
300 — ‘Angl-Fr’
400
500
600 OHG
700 OE Os
800 OLF
900
E

1000 M MLG MDu MHG~ OYi
1100

1200 OF ri

1300

1400

1500

1600

1700 ModE Fri LG Du Afr G Yi

Pict RE 7. A detailed history (relatively un-generalized) of the West Germamie languages, after Barber (1993). The terminal branches shown
are Modern English (ModE), Frisian (Pri). Low German (LG), Dutch (Du), Afrikaans (Afr), German (G), and Yiddish (Y1). This diagram is a
more highly resolved representation of the lower right branch (WGme) in Figure 6

(oa)

TEMPORAL REPRESENTATION

991] DY] SSOIOK S]UDWIA[A aSeNSUP] JO BUIMOLIOG OsTe ING UOISSILUSURI} JIOIIP A[UO OU

SMOYS 391] SIU] “UOISSILUSUB.) 9911p ATUO 1Idap YotyAr “7 puke g SANS O} ISBINUOS UY “(YRG]) [[OMog Joye ‘sasensuer] uvadoiny-opuy paysajas Jo ABojwauad y “8 ANNI]
YyOoD)
ysj|Guyq uwsapow = youasj usapow UJapOW ABD JUBSAld
> el Oe ae —-—--—--— — — — — — [sabeuloo oinUa!9sS]
Pei Se ST RCRR Stay alae reer ee M
usi|buy wapoyw Ayeg youas4 wepoy Aye3 Rear ar tA
| \
ysi6uz aIPPIN<— —_ _ _ @—--—--—--—--—---— ~~~. — \ av 00S}
< — — — youas4 UBWION ;
sabenbue| \
BOUBWOY \
[puejbuz 40 jsanbuog iad <_ youedd PIO Jayjo \ av 0001
(RISING wetercat el oo ee ol i ee ee SS \
uoxes-o|buy 77> = — ~ aouewoy Aiea \
\
ee ee oe, auljuezAg
[ulejug 0} ayes6iw uoxesg-ojbuy] => — —uneqsebina ule] usu \ av 008
Sa a a (uayods) < \
NOt ee a a Se ~
sabenbue| uoxes ysuueg = (atsti(‘<‘é‘a‘C;OC*drSO!!!UUUUU — — —yoinug ueysuyg
DIUBLWIB®) JOYJO jejuaul}uOD ‘14J0D * JO 48015) av
uneT 1ed1SseIQ <— _ \
$}9/BIp No SS : og
oul] ee Ea
434}0 a in Pelee 42949 ene
~ 2. = ~~ a i \
$J94yjO d10q oN Og 00S
s9leIP 4895
| Od 0001
SAYIUEIG JAYJO QIUBWIAD uel IIUA/|OH
Od 0002

ueeadoinz-opu|

TIME
IN
M.Y.B.P.
(MA)

1.0

Tad

1.4

1.6

1.8

—p 2.0

3.0

4.0

_p 5.0

FIGURE 9. A portion of a geological time scale, redrawn after Haq
and Van Eysinga (1987). The linear scale stretches out toward the
present so that more events can be included. This stretching of the
temporal space is equivalent to the practice of feature displacement in
cartography, but when it 1s carried to this extreme in cartography the
effect is usually regarded as humorous

ROBERT J. OHARA

Again in the spirit of palaetiology, I offer a linguistic ex-
ample to show that the same principles apply once again. In
the previous diagrams of Germanic language phylogeny
(Figs. 6 and 7) the pattern of descent was entirely vertical,
and it gave the impression that each language evolved inde-
pendently and in isolation from its sister languages. From
Figure 8, however, a diagram showing the history of selected
Indo-European languages (Powell, 1988), we get a very dif-
ferent sense of the growth of Modern English. While there
is indeed a line of transmission coming down from the early
Germanic languages to Modern English, the ancestors of
English are seen here to have received elements from a va-
riety of sources, including Greek, Latin, and French. Does
the fact that none of this linguistic borrowing is shown in
Figure 7 mean that Figure 7 1s false? Not at all: Figure 7
correctly depicts certain classes of events, while Figure 8
depicts many of the same events as well as some additional
events. The relations among these diagrams are conceptually
identical to relations that can be observed in cartography,
for example between a map that shows a number of highways
running in parallel, and another map that shows not only
those highways but also a network of small roads that connect
them.

One of the cartographic phenomena | mentioned above
was feature displacement, a local warping of the scale that
occurs when two objects are nudged closer together or farther
apart in order to accomodate the desire of the map-maker
to include a certain collection of map elements. When this
is done to a limited extent it isn°t noticed, but it can be
concentrated for special humorous effect, as in the various
entertaining maps that illustrate “A Bostonian’s View of the
World” or “A New Yorker's View of the World” (Gould
and White, 1986). Is there a temporal equivalent of this warp-
ing of geographical space? There is, and it can be seen in
at least two different palaetiological contexts. The first 1s in
phylogenetic trees that stretch out around particular taxa,
most often humans, and which thereby create “A Human’s
View of Evolutionary History” that is conceptually identical
to maps showing “A Bostonian’s View of the World”
(O'Hara, 1992). A second palaetiological context in which
the temporal equivalent of feature displacement can be seen
appears in Figure 9, which reproduces a portion of a widely-
used chart of geological time (Haq and Van Eysinga, 1987)
on which the temporal scale changes repeatedly. The design-
ers of this chart wanted to fit more temporal detail into the
time scale in more recent periods, and to do so had to stretch
out the temporal space. Once again, this process 1s concep-
tually identical to the warping of geographical space that we
see in feature displacement, but it is carried here to an ex-
treme that in cartography would be regarded as consciously
humorous. It 1s worthwhile to consider how such warping
of temporal space affects our sense, and particularly our stu-
dents’ senses, of evolutionary time and the history of the
earth.

Let me close by suggesting a way in which this last ques-
tion — the effect of conventional patterns of temporal gen-
eralization on students’ perceptions of evolutionary history
— might be addressed. Geographers have done quite a bit

TEMPORAL REPRESENTATION

se ay d Neha '

\
acthro Las

\lusks P
oi ally fish

com 5 puvaes

unicellular Cvkera itt)

Jal cu ulue prokaryotes

FIGURE 10. An evolutionary tree drawn by an undergraduate biology student at the University of Wisconsin. On the first day of a course
on evolution each student was given a sheet of paper and was instructed to “sketch an evolutionary tree of life, and label as many branches as
you can. Don’t worry if your tree is not perfect, or if you can’t remember technical terminology; this is not a graded exercise. and you should
not even put your name on the page.” Exercises such as this, which are modelled on geographers’ studies of “mental maps” (Gould and White,
1986; Saarinen, 1988; Walmsley et al., 1990), may help evolutionary biologists to better understand popular conceptions of the history of life

and to develop more effective teaching strategies.

thee A meres Misms

Anna ( binqilon.

_ “ les vorlebe ale 5

rl wit insects

sets» h achnide

FIGURE 11. A second student-drawn evolutionary tree.

of research on what are called “mental maps” (Downs and
Stea, 1973; Stevens and Coupe, 1978; Gould and White,
1986: Saarinen, 1988; Walmsley et al., 1990). If we take
any person and ask him to draw from memory a map of his
hometown, or of the world, or of any other region, the re-
sulting map will reveal a great deal about that person’s
knowledge of geography, his perception of the sizes and dis-
tances between various geographical objects, and so on. Is
it possible to do this same sort of research with historical
representations? It is, and I offer here two sample results
(Figs. 10 and 11) from a preliminary inquiry of this type,
carried out with the assistance of Gregory Mayer at the Uni-
versity of Wisconsin. Students in a large undergraduate
course on evolution were asked on the first day of class to
draw an evolutionary tree of life as best they could, based
on whatever knowledge they may have acquired from general
reading or from other courses they may have taken. A great
variety of images were produced by the students in this ex-
ercise. Many of them have a decided axis that leads to human
beings, suggesting that there 1s still a widespread belief that
evolutionary history is progressive or directed (O'Hara,
1992). A number of the diagrams clearly reflect the “five
kingdom” arrangement of Margulis and Schwartz (1988),
something that many of the students appear to have been
taught in secondary school. Very few of the diagrams would
be regarded by a contemporary systematist as particularly
accurate. | hope to extend this preliminary study to other
groups of students at other institutions in the future, and
thereby build up a general picture of undergraduate under-
standing of evolutionary history.

William Whewell, with whom I began this essay, was not
only an historian, a philosopher, and a scientist, he was also
an educator: he served for many years as Master of Trinity
College in Cambridge, and wrote university textbooks as well
as essays on the importance of liberal education. Whewell
believed that the palaetiological sciences — the historical sci-
ences — were particularly well-suited for inclusion in a gen-
eral liberal curriculum because they exemplify not only rig-
orous forms of thought and argument, but also the enormous
reach of human reason by taking us farther back in time
than previous generations of scholars would have ever
thought possible. I think Whewell was right. I also think that
by strengthening the ties that bind together all the historical
sciences
time — we will in turn strengthen our own particular special
fields, and will be able to do a better job of explaining our-
selves to our colleagues and our students in the future.

Acknowledgments

lam very grateful to Michael Ghiselin and Giovanni Pinna
for giving me the opportunity to participate in their symposia
on historical biology at Milan and San Francisco. Jeffrey
Wills and Peter Robinson contributed greatly to my under-
standing of philology, Gregory Mayer assisted with the
“mental tree” project, and Laurie White offered valuable
comments on the manuscript. [am also grateful to the many
members of my network discussion group Darwin-L, who

all the disciplines that try to map the space of

ROBERT J. OHARA

have done much to enlarge my understanding of the historical
sciences.

Literature Cited

Auroux, S. 1990. Representation and the place of linguistic change
before comparative grammar. Pages 213-238 in T. De Mauro
and L. Formigari, editors. Leibniz, Humboldt, and the Origins of
Comparativism. John Benjamins, Amsterdam.

Barber, C. L. 1993, The English Language: a Historical Introduc-
tion, Cambridge University Press, Cambridge.

Buttenfield, B. P., and R. B. McMaster, editors. 1991, Map Gener-
alization: Making Rules for Knowledge Representation. Longman
Scientific and Technical, Harlow.

Celakovsky, F. L. 1853. Readings on Comparative Slavic Grammar
at Prague University. F. Rivnac, Prague. (In Czech)

Collin, H. S., and C. J. Schlyter. 1827. Corpus Juris Sueo-Gotorum
Antiqui. Volume I. Z. Haeggstrom, Stockholm.

Crampton, J, 1990. An elusive reference: the 1:1 map story. Car-
tographic Perspectives 8:26-27.

Darwin, C. 1859. On the origin of species by means of natural
selection, or the preservation of favoured races in the struggle
for life, Ist Edition. John Murray, London,

de Queiroz, K. 1988. Systematics and the Darwinian revolution.
Philosophy of Science 55:238-259.

Downs, R. M., and D. Stea, 1973. Image and Environment: Cog-
nitive Mapping and Spatial Behavior. Aldine Publishing Com-
pany, Chicago.

Ghiselin, M. T. 1991. Classical and molecular phylogenetics. Bollet-
tino di Zoologia 58:289-294.

Gould, P., and R. White. 1986. Mental Maps, 2nd Edition. Allen
and Unwin, Boston,

Haq, B. U., and F. W. B. Van Eysinga. 1987. Geological Time Ta-
ble, 4th Edition. Elsevier Science Publishers, Amsterdam.

Hennig, W. 1966. Phylogenetic Systematics, 2nd Edition. Univer-
sity of Illinois Press, Urbana, Translated by D. Dwight Davis
and Rainer Zangerl

Holm, G. 1972. Carl Johan Schlyter and textual scholarship. Kun-
gliga Gustav Adolf Akademiens Aarsbok 1972:48-80.

Maddison, W. P., and D. R. Maddison. 1992, MacClade, Version
3. Sinauer, Sunderland.

Margulis, L., and K. V. Schwartz. 1988. Five Kingdoms: an IIlus-
trated Guide to the Phyla of Life on Earth. W. H. Freeman, New
York.

McMaster, R. B., and K. S. Shea. 1992. Generalization in Digital
Cartography. American Association of Geographers, Washington,

TEMPORAL REPRESENTATION

Monmonier, M. 1991. How to Lie with Maps. University of Chicago
Press, Chicago.

O'Hara, R. J. 1988a. Diagrammatic classification of birds, 1819-
1901: views of the natural system in 19th-century British orni-
thology. Pages 2746-2759 in H. Ouellet, editor. Acta XIX Con-
gressus Internationalis Ornithologici. National
Natural Science, Ottawa.

———. 1988b. Homage to Clio, or, toward an historical philosophy
for evolutionary biology. Systematic Zoology 37:142-155.

. 1991. Representations of the natural system in the nine-
teenth century. Biology and Philosophy 6:255-274.

1992. Telling the tree: narrative representation and the
study of evolutionary history. Biology and Philosophy 7:135-160.

———, 1993. Systematic generalization, historical fate, and the
species problem. Systematic Biology 42:231-246.

Powell, J. G. F. 1988. Introduction to Philology for the Classical
Teacher. Joint Association of Classical Teachers, London.

Priestly, T. M. S. 1975. Schleicher, Celakovsky, and the family-tree
diagram. Historiographia Linguistica 2:299-333.

Museum of

eed

Robinson, A. H., and B. B. Petchenik. 1976. The Nature of Maps
University of Chicago Press, Chicago.

Saarinen, T. F. 1988. Centering of mental maps of the world. Na-
tional Geographic Research 4:112-127.

Sober, E. 1988. Reconstructing the Past: Parsimony, Evolution, and
Inference. MIT Press, Cambridge.

Stevens, A., and P. Coupe. 1978. Distortions in judged spatial re-
lations. Cognitive Psychology 10:422-437.

Toulmin, S. 1953. The Philosophy of Science: an Introduction.
Hutchinson University Library, London.

Tufte, E.R. 1983. The Visual Display of Quantitative Information.
Graphics Press, Cheshire.

Walmsley, D. J., T. F. Saarinen, and C. L. MacCabe. 1990. Down
under or center stage? The world images of Australian students.
Australian Geographer 21:164-170.

Whewell, W. 1847. The Philosophy of the Inductive Sciences, 2nd
Edition. John W. Parker, London.

PERIODIZATION AND MODELS IN HISTORICAL BIOLOGY

James R. Griesemer

Wissenschaftskolleg zu Berlin
and
Collegium Budapest

Permanent address: History and Philosophy of Science Program
University of California, Davis California 95616-8673

The nourishing fruit of the historically understood
contains time as a precious but tasteless seed.
Walter Benjamin

1. Introduction

In a previous paper I argued that what makes a science
or a scientific theory historical is that its practitioners accept
historical narratives as informative (Griesemer, 1996).' The
informativeness of narratives lies in their contributions of
data, explanations, and interpretations to understanding. This
view emphasizes a pragmatic aspect of historiography in
contrast to those that emphasize either alleged intrinsic his-
torical properties of the objects of study or an epistemologi-
cally privileged standpoint of historical explanation.” One
need not take such an ontological or epistemic view to find
something importantly distinctive about historical science.
Since part of my goal was to explore reasons scientists call
a theory or science historical and how historical scientists
defend themselves against critics, rejection of these com-
monplace readings of historical science led me to consider
pragmatic aspects of historical science.

Given the pragmatic view that a science is historical be-
cause its practitioners treat it as such, | here develop the
basis for a line of defense of historical sciences against the
charge that they are non-scientific or pseudo-scientific. Evo-
lutionary biologists are often lampooned for offering “just
so” stories in place of empirically grounded, testable, mecha-
nistic explanations. Systematists, paleontologists, and bio-ge-
ographers are criticized for offering “pattern” descriptions
without “process” explanations, while historical scientists
complain that process — in the form of mechanistic models
—is routinely injected into historical sciences without bene-
fit of a broad perspective on either the phenomena or the
interpretive problems. The message of this paper is that the
usual caricature of narrative science that supposedly warrants
such derision underestimates or entirely misses the theoreti-
cal character and import of narrative science.

The charge against historical science rests on the argument
that it consists in the “Jjust-so” stories, rather than testable
explanations in terms of mechanistic models, but only the
latter support properly theoretical science. So historical sci-
ence is really not scientific. Two lines of defense are common
(see Griesemer, 1996), One says that historical science is

New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

scientific because not all (good) science relies on mechanistic
models; some of it is narrative. The other says that the “real”
science in historical science isn’t really narrative, it is mecha-
nistic just like physics. Neither line of defense gets to the
heart of historical science, which is narrative but nor anti-
mechanist. The historical sciences derive their distinctive
characteristic form not from the intrinsic nature of their sub-
ject matter, nor from pure narrative, but from theoretical
coordination of both mechanism and narrative. So a better
defense is first to identify the respectable theoretical activity
associated with narrative construction and then to argue that
this activity is not inimical to good mechanistic modeling
and explanation, thus showing that historical sciences can
be judged favorably on the same grounds as other sciences
(e.g., the robustness of their models), even while taking nar-
rative seriously.

The goal of this paper is thus to reveal, illustrate and ex-
amine historical scientists” engagement in theoretical model
construction and evaluation. The illustrations chosen are not
exhaustive, nor do they provide an overview of the historical
sciences. Rather, they attempt to demonstrate that certain
issues arise In a variety of modeling circumstances as a con-
sequence of the need to choose a temporal framework for a
narrative. The examples have been drawn from the emerging
intersection of evolution and development.

Scientific modeling depends on establishing a “state space
in which state variables, parameters and laws are well de-
fined” (Lloyd, 1988). An important element of the framework
for choosing a state space type? is what I call a “periodiza-
tion.” Since the formulation of alternative models in histort-
cal biology can be linked to distinct modes of adequate me-
chanical description within the historical periods of a
periodization, narratives can be understood as coordinators
of mechanical models. Population genetic models, for exam-
ple, appeal to mechanisms underlying gene transmission, but
descriptions of the temporal structure of transmission require
choice of a temporal framework. Narratives of genetic events
are not mere stories, creations of the narrator without em-
pirical grounding.

Narratives are not usually considered to be founded on
robustness analysis of alternative models and modeling
frameworks; one simply tells a story. But narrative structure
depends on the choice of time scale and temporal units such
that mechanical descriptions within and among time periods
are coherent and robust. So narratives are not singular Just-so

stories indicating how possibly events became ordered in a
particular sequence. Rather, they have a more complex theo-
retical structure. The robustness of the underlying mechanis-
tic models depends, in some measure, on the periodization
selected as a framework. But then it becomes important in
evaluating narratives to assess how well the selected frame-
work supports empirically adequate models of phenomena
linked in the narrative. Because evaluation is comparative,
it requires consideration of alternative periodizations. These
facts entitle one to say that narrative science 1s deeply theo-
retical.

While evolutionary narratives have been singled out in
criticisms of historical science, | believe that a thorough un-
derstanding of historical biology should take account of de-
velopment as well as of genealogy. Although some historical
work has been done on special problems about the relation-
ship between evolution and development, e.g., Gould’s study
of the recapitulation principle (Gould, 1977), less has been
done to explore the ramifications for historical science of
the general philosophical relationships between evolution and
development, save for Ghiselin’s penetrating study of Darwin
(Ghiselin, 1969). Vermeij (1987) offers a valuable analysis
of the hypothesis-forming role of an ecological theory of
adaptation for historical inquiry into evolutionary biology.
Consideration of relationships between evolution and devel-
opment can be used to illustrate the coordinating role of
periodizations by examining how developmental mechanism
plays a role within evolutionary narratives. This is not, of
course, the only, or even the most interesting case of coor-
dination, but it is an important one given current moves to
integrate evolutionary and developmental theory.

To the standpoint that historical understanding does not
lie at the opposing pole of a dichotomy with mechanism, |
want to add that mechanical and narrative accounts are both
facilitated, and indeed linked, by the use of periodizations,
This conclusion can be used to defend historical science on
the grounds that good history does not in fact do without
mechanism; critics and supporters of historical science share
a misunderstanding when they suppose it to be purely or
solely narrative in form,

In section 2, | discuss the idea of periodization and some
reasons for entertaining it in regard to the historical sciences.
In section 3, [ argue that an important role of periodizations
depends on their abstract and hypothetical nature, which
makes consideration of a/ternative periodizations a flexible
and powerful methodological tool. In section 4, I discuss
several examples to show how choice of periodization may
facilitate homogeneous, mechanistic explanations within
stages and may add to our understanding of the relationship
between evolution and development. Section 5 draws some
conclusions for the prospects of a defense of historical sci-
ence from these arguments and illustrations.

2. The Concept of Periodization
My thesis is that there is a central theoretical structure

common to all historical work which I call a “periodization.”
A periodization is a conventional marking of time into parts

JAMES R. GRIESEMER

or stages. In many sciences, periodization is given by “New-
tonian time” and very little explicit attention is given to its
theoretical role. In other sciences, typically those involving
“stage” theories, the duration and structure of the stages or
periods is a focus of concern, partly because a poor choice
of units leads to difficulty in describing transitions among
stages. Periodization plays a more or less prominent role in
historical writing: in some cases it amounts to no more than
the organization of an annal or chronicle by calendar years
or clock minutes, but in others a new periodization may
change dramatically what we count as historical events and
how we think about their significance. An annal is a simple
listing of events in chronological sequence without interpret-
ing their significance. A chronicle is an interpretation of a
sequence of events that is simply cut off at some point (often
the writer’s present) without any assessment of significance
or meaning — an unfinished story. A narrative concludes a
chronicled story with an interpretation of the significance of
the events selected for inclusion in the story.°

Periodization challenges how we understand events. In hu-
man history, for example, questioning [8th—20th century pe-
riodizations of the 16th and 17th centuries that were based
on Enlightenment conceptions of intellectual progress, brings
into doubt the nature of “The Scientific Revolution” as an
historical event (Dobbs, 1994). Similar things have been done
to “The French Revolution” and to the concept of revolution
in general (Latour, 1993). Of course, the most obvious ex-
amples from natural science of how a change of periodization
may lead to a changed understanding of events come from
geology and paleontology.

Historical studies of how Victorian geologists came to rec-
ognize the ancient origins of life through their arguments for
the existence of the Cambrian, Silurian and Devonian periods
provide particularly good examples: by arguing for distinct,
early fossil-bearing strata, geologists pushed back the ap-
pearance of life in time and revised their understanding of
fossil and stratigrapic sequences.° In particular, debate about
the use of lithological vs. paleontological criteria for estab-
lishing the boundaries of strata and geological periods re-
flects a changing understanding of the relationship between
mechanical description — literally how the rocks are struc-
tured and how they got that way — and periodization, how
the choice of temporal units in which to describe fossil and
rock sequences constrains claims about the existence of geo-
logical periods. Although outside the scope of the present
paper, application of the ideas developed here to geological
examples would be instructive,

Periodization of time is often inextricably tied to region-
alization of space, that is, a division of space into regions
that identify what counts as a “local” interaction and what
would count as “action at a distance.” In biology, “individu-
ating” processes such as cell division, organismal reproduc-
tion and cladogenetic speciation, which all lead to the crea-
tion of new individuals (cells, organisms, species), require
that participating entities be in temporal and spatial proximity
in very specific senses. A description of the sequence of
events of a biological process such as allopatric speciation
involves both spatial and temporal sequences. A description

PERIODIZATION AND MODELS

of the temporal stages of the cladogenetic process underlying
speciation entails a regionalization of populations, for exam-
ple, into central and peripheral components where founding
isolates are formed according to the Mayrian allopatric
model. Because of the intimate relationship between tempo-
ral and spatial locality with respect to a given mechanical
process, temporal descriptions can represent spatial regions
and conversely.

A periodization can place constraints on spatial relations
admissible in a model of an individuating process. Con-
versely, a choice of spatial relations can constrain the ad-
missible periodization of events described by a model. In
embryology for example, spatial relations can reflect tempo-
ral sequences: spatially separated sister-cells must, in general,
have undergone a certain time-evolution in order to have
become spatially separated since the cell-division event that
created them. If two cells are next to each other at a given
time, but not afterward, then it is likely that they had a com-
mon ancestor. This is true to the extent that relative cell
movement results from cell division. Patterns of spatial sepa-
ration of genealogically related elements can thus be used
to represent temporal stages. This is common in classical
descriptions of developmental stages: e.g., the “late neurula
stage of frog development” might indicate a certain fime in
development even though the stage itself is characterized
primarily in terms of the spatial pattern in which cells on
opposite sides of the neural fold are in contact.’

Minelli suggests that complexity principles used to inter-
pret arthropod segmentation in terms of tagmata (body re-
gions) may apply to the time dimension as well, “. . . if
developmental stages are equated to body segments and ma-
jor developmental phases (such as insect larva, pupa, and
imago) are equated to body regions.”* He argues that a causal
link has already been established by molecular biologists be-
tween the temporal dimension of development and the spatial
dimension of the longitudinal body axis, citing the pattern
of expression of Hox genes.” Spatiotemporal unity and con-
tinuity are properties that “integrate” individuals into histor-
cal entities (Hull, 1975). Models that rely on an explicit di-
vision of time into units may help identify or define such
entities. A periodization provides the framework of temporal
units in which that integration (and its mechanisms) can be
described —a “state space” in formal modeling terms.

Historical description is facilitated by periodization to the
extent that events and objects within periods can be described
according to a single perspective on, or “mechanistic” ac-
count of, causal processes. The unified perspective that a
periodization provides can be characterized as the framework
for choosing a state space in which to describe a theoretical
model, according to the semantic view of biological theories
(Lloyd, 1988). Models are structures that satisfy the law-
statements of scientific theories. Theoretical models are ab-
stract structures that make such laws true, and when hy-
pothesized relationships between theoretical models and
empirical structures in nature hold, the laws are true of nature
as well (Giere, 1988). Theoretical models thus form the mid-
dle term of a relation between laws, models and structures
identifiable in nature. Tests of hypotheses require comparing

structure observed in nature with the structure of models. To
the extent that model structure depends on choice of a state
space, periodization is implicated in the conditions for testing
as well as constructing hypotheses.

Periodizations serve as the framework in which models
classify objects and events through assignment to a particular
unit of time. The assignment reflects laws of coexistence!”
and succession,'! thus there is an intimate relationship be-
tween the form of the (possibly implicit) laws of a theory
and a periodization. Natural historians of all sorts make and
use such classifications, a fact that extends the reach of the
historical to most “mechanistic” fields in biology such as
genetics, where time is divided into generations, and bio-
chemistry, where time is divided into reaction cycles. Divi-
sion of time into stages affords a mode of classification that
plays a wide theoretical role in historical science. The ways
that temporal classifications articulate with spatial, structural,
or genealogical classifications are complex and cannot be
pursued further here.

It is advantageous to think of periodizations as frameworks
for specifying a state space of models to emphasize that the
historian may entertain alternative periodizations for a given
set of models, objects or events, on the one hand, and to
entertain alternative models (or state spaces) within a given
periodization on the other hand. Models are cheap while data
are dear, and it behooves the careful investigator to explore
alternative interpretations in order to make the most of data.
One way the consideration of alternative models does this
is by guiding the search for robust empirical consequences
in which the fit between models and data is shown to be
independent of the artificial “simplifying” assumptions of
any particular model (Levins, 1966; Wimsatt, 1981, 1987).
“Robustness analysis” of scientific models requires more
than one model of a phenomenon for there to be an adequate
scientific account of it. By the same token, the robustness
of a collection of models to a change of periodization may
reveal important generalities about a phenomenon.

Given a conception of narrative pure and simple, the role
of alternative models in historical science 1s easily over-
looked. A second story does not seem necessary as a “reality
check” on a given story. One just tells the story (well or
badly) and moves on. But checking the fit of a model to a
selected periodization and considering alternative periodiza-
tions are common activities in the historical sciences. How-
ever these activities are not particularly evident to the casual
observer nor are they typically interpreted in the spirit of
modeling that is so common in mechanistic sciences. Iden-
tifying periodization as an important theoretical activity
should help draw attention to this important aspect of his-
torical science.

The most important properties of adequate periodizations
are that they support a unified description of events within
stages and that they identify breaks or discontinuities be-
tween stages. Minelli suggests that the changes occuring
within molts of post-embryonic arthropods are “. . . the de-
velopmental equivalent of the smooth gradual changes in
morphology we can observe with one tagma, along the an-
tero-posterior axis of the body.”!* Other more dramatic

changes or discontinuities are the metamorphoses that con-
stitute major developmental events. Both are critical to the
mechanistic analysis of the existence and persistence of
(theoretical) individuals. In some sense, a gradual tempo and
a continuous mode are hallmarks of (simple) biological
mechanism. In many historical sciences, all the events within
a given stage of a periodization are subject to investigation
according to a single developmental-mechanical perspective.
That is, the function of periodization is to divide time into
stages that are explanatorily homogeneous, and within which
a single, ahistorical mode of description can be employed
by a single model type. Within-stage events are accounted
for as the result of causal processes that can be given non-
narrative, mechanical descriptions. For those of a certain
bent, stage theories can be considered place-holders until a
single, unifying mechanism can be found that makes the
whole process explanatorily homogeneous. Transitions be-
tween stages are usually poorly explained compared to se-
quences of events within stages, but narratives string stages
together to make some sense out of a whole process and
pave the way for more detailed attack on the transitions.'*

Relative to a given modeling purpose, an adequate perio-
dization will serve to single out, or at least to order, subsets
of available mechanical principles with which to interpret
within-stage events. In historical biology, embryology is the
premier example of the mechanical study of changes in or-
ganisms that feeds into the narrative account of evolutionary
change in populations, species and higher taxa, Organisms
develop so that they reach reproductive condition. The re-
productive nexus is what organizes populations genealogi-
cally. So the events that take populations from one organ-
ism-generation to the next are describable in terms of the
mechanisms of development.

Mechanisms at the higher levels are often functions of
those at lower levels (as for example organismal development
is a function of migration of cells), so the links between
embryology and evolution are complex and neither is dis-
tinguishable as the purely mechanistic or the narrative part-
ner. An evolution-development narrative would not follow
a single “central subject” or individual up the levels of or-
ganization, for example, from cell to organism to population
to clade. Rather the description of dynamics of individuals
at a given level would be linked to individuals at other levels
by showing how a choice of time scale appropriate to one
level would constrain characteristics of appropriate models
at other levels. One example of this is the way geneticists
describe life cycles. Because the continuity that population
geneticists need is trans-generational, the only cellular events
that really matter (in classical models) are the ones leading
from zygote to adult to new gametes. The within-generation
continuity of the “phenotype” is irrelevant. So a trans-gen-
erational periodization for genetical purposes imposes a pe-
riodization on embryological events that 1s coarse-grained
from the embryological point of view for the sake of a ho-
mogeneous narration of genetical processes.

The mechanical description of embryological events may
differ among embryological stages, however, and the coarse-
grained periodization of cellular events by the geneticist may

JAMES R. GRIESEMER

not be adequate for other theoretical purposes. The earliest
events of metazoan development, for example, are usually
described in sub-cellular, biochemical terms relating the po-
sitioning and movement of molecules within a single cell.
Somewhat later events (at a time when the organism ts still
a multi-cell) are described in terms of cell-cell interactions.
Still later events are described in terms of germ-layer or
tissue interactions and movements. Each of these sets of
events may occur on a different time-scale. But successful
periodization only requires unity of the mechanical form of
description within a stage; it does not require that all stages
be pursued with the same form or style of description, or
that stages be of the same duration. Periodization therefore
does not assume any form of reduction to the lowest common
physical or chemical denominator in linking together stages
in a narrative. Nor does the methodological unity suggested
by the practice of periodization suggest any direct linkage
among phenomena in different fields such that there can be
theoretical unification of the sort promised by the modern
evolutionary synthesis (linking genetics, paleontology, de-
velopment, ecology, and systematics). It is not obvious that
narratives constructed from different theoretical perspectives
can be unified. We do not need to claim this much, however,
to argue the point that a given narrative from a particular
theoretical point of view can serve the constructive purposes
I have been describing. It would be interesting, but far be-
yond the scope of this paper, to explore the possibility that
apparently incompatible narratives such as the embryol-
ogist’s and the geneticist’s alluded to above could flow to-
gether to form new (probably multi-level) theoretical entities.

Causal-explanatory unity within each stage is a necessary
but not sufficient condition of adequacy for a theoretical his-
torical model. To be adequate, a model must also support a
narrative that links stages. Hull (1975:198) has emphasized
that narratives do this primarily by describing a “central sub-

ject.” a historical entity that functions as the core of a nar-

rative. In addition to dividing time into units within which
events can be given unified mechanical explanations, perio-
dizations set the framework for identifying the integration
of individuals into historical entities capable of serving as
central subjects. In biology, most individuals at a given level
of organization are composed of individuals at lower levels.
Multicellular organisms are historical individuals that are
composed of cells, which are themselves historical individu-
als at a lower level, composed in turn of organelles and mole-
cules that are historical individuals at still lower levels. By
setting the “break points” in a sequence of stages within
individuals at a given level and among individuals at lower
levels, a periodization constrains the selection of the genea-
logical links to describe the integration of lower-level indi-
viduals into a higher-level individual. Periodization thus
plays a role in the identification and individuation of his-
torical individuals.

Historical description involves postulating (or assuming)
a rough periodization of the temporal phenomenon of interest
based on prior orderings of events into chronological se-
quence (e.g., a division of time into seconds, minutes, hours,
days, years, seasons, generations). Periodization can also be

PERIODIZATION AND MODELS

implicit in the formation of a preliminary historical hypothe-
sis to the effect that “events occur in chunks like these.” In
historical research one roughs out an “annal,” which is a list
of events ordered in chronological sequence according to the
time units of the periodization, and a “chronicle,” which is
a temporal sequence of events at different times within stages
leading to the culminating states at the end of each stage.
Events and objects within stages must be consistent with
principles of a mechanics adopted for each stage. One at-
tempts to establish narrative connections within and among
stages in the model that constitutes the historical explanation.
At its crudest, a narrative will relate one event from each of
two different stages and be “about,” 1.e. interpret, the earlier
event “in light of” the later one (Danto, 1985).

It is important to see that a periodization model must be
constructed prior to the firm establishment even of a chron-
icle. Without periodization, there is, so to speak, no way to
select and describe events historically. This will be so re-
gardless of whether history must be narrative or could include
pure annal or chronicle forms as genuinely historical (cf.
White, 1987). In the context of narrative history, the perio-
dization coordinates mechanical explanation of events with
narrative description of a central subject by simultaneously
facilitating the identification and selection of events to be
included in a narrative and the reification of an individual
into a historical entity worthy of being a central subject in
a narrative. Without a periodization, not only could there
not be a narrative, but there could be no subject of narration.
Hull expresses well the dual requirements served by perio-
dization models when he writes that,

two sorts of linkage are involved in historical
narratives: one the cause-effect relation connecting the
events associated with the historical entity, and the
other the part-whole relation integrating the central
subject into a single historical entity (Hull, 1975:187).

These unified, within-stage cause-effect or mechanical de-
scriptions will tend to look rather different than the overall
coordinated accounts that describe meaningful sequences
among stages because the latter tend to be narrative in form
while within-stage descriptions will be causal-mechanical.
Although Hull (1975:196) characterizes narrative as“... a
description of the central subject and the events in which
this subject participates,” Danto’s point, that narrative sen-
tences have a distinctive logical form because they refer to
events later than an event of interest but are only “about”
the earlier one, must be added to clarify the contrast between
mechanical and narrative description (Danto, 1985; see also
Griesemer, 1996). This point is critical to understanding his-
torical science, because there is a tendency to identify it only
with narrative activity and not with the mechanical analysis.
But both are critical to historical analysis, and this is more
clearly seen through consideration of the central coordinating
role of periodization as theoretical model.

3. Periodization and Alternative Models

If the modeling function of periodization is common to
all biological sciences, then historical biology will always
bear an important relation to ahistorical biology. In Ernst
Mayr’s (1961) apt terms, ultimate biology (evolution) will
always be tethered to some proximate biological theory (e.¢.,
in physiology, genetics, morphology, cell biology, biochem-
istry) that provides the basis from which to produce the me-
chanical descriptions of events within periods that are incor-
porated into evolutionary narrative.

A periodization frames a collection of models in the sense
that it defines a space for state descriptions of objects of
interest and the events that involve changes in state. The
laws of transformation that describe such changes (causes,
human actions) are contained in the mechanical theories that
apply within stages. As I said above, models are abstract,
hypothetical structures. They are applied in the form of hy-
potheses that the structure of nature conforms to the structure
of the model. Most effective science involves formulation
of a variety of models that are intended to represent different
aspects of the phenomena under study rather than treating
the abstract framework of a given model as a true description
of nature simpliciter. The art in scientific modeling 1s choos-
ing a set of models whose artificialities are independent of
one another, so that “our truth,” as Richard Levins put it,
“is the intersection of independent lies” (Levins, 1966:423).

Much as under ordinary circumstances there are no fric-
tionless planes, the beginnings or endings of decades or cen-
turies typically do not mark the most significant historical
events just because they are nice round numbers. But the
division of time into centuries plays a role in history analo-
gous to that of the frictionless plane in mechanics. It is an
idealization that facilitates representation of complex expe-
rience, one that helps us identify the mechanisms that explain
events and also describe the entities (historical individuals)
that participate in those events. Although it may be a false
hope that the same set of “forces” that accounts for political
events at the beginning of a century will account for events
at its end, a one-century periodization is more likely to suc-
ceed than a two-century periodization. Here one can envision
the trade-off between model “realism” and “precision” that
Levins wrote about. An ultimately realistic periodization
would have “tick marks” at the same temporal scale as the
(lowest-level) individual events, but such a model would be
as worthless for purposes of representation and understanding
as to treat a cat as if it were a model of a cat. In historical
description, even the division of time into segments of ar-
bitrary size allows one to entertain the possibility and fea-
sibility (given current mechanistic understanding) of uniform
description of events within segments and to describe the
continuity of individuals among segments.

Notice that the concept of periodization introduces a com-
plication into the contrast between annal or chronicle, on the
one hand, and narrative on the other hand. A periodization
imposes a structure on the ordered, but not necessarily scaled,
elements of a chronicle. But it does not fully construct the

complex relations between future and past relative to a given
event or time that are characteristic of a narrative. A perio-
dization coordinates the relationship between a narrative and
the data contained in historical documents by constraining
both. The periods structure our views of the data as repre-
sented in models and at the same time organize the terms
of the narrative. Periodizations, in other words, coordinate
the “empirical substructures” specified by chronicles within
the framework of a narrative account and stand in analogous
relations to chronicle and narrative that models do to phe-
nomena and theory.'4

The view that historical interpretation involves postulating
a framework for theoretical models allows me to express a
significant, under-explored problem in the analysis of evo-
lution as a historical science: the relationship between evo-
lution and (embryological) development. | want to consider
some examples from evolutionary developmental biology to
illustrate the modeling role of periodization. A detailed
analysis or case study of the articulation of evolution and
development would be premature. Here, the idea is simply
that periodization of development establishes the framework
for mechanical explanation of evolutionarily significant
events in which organisms participate. | will assume that
organisms constitute historical individuals that are parts of
species, which are themselves historical individuals. The
main point to be illustrated is that a change in the periodi-
zation of development can have substantial consequences for
the construction of a historical evolutionary narrative about
species or other taxa. This is obviously not a new idea. One
could interpret the whole history of the recapitulation concept
and biogenetic laws as a matter of the proper periodization
and registration of phylogeny and embryology (Gerson MS,
ch. 2). What is new is the idea that the periodization of
development constrains the structure of theoretical historical
models in this field.

Let us consider an illustration of how changing periodiza-
tion so as to more adequately capture mechanical descriptions
of within-stage events may lead to alternative evolutionary
narratives. Critiques of historical science that appeal to the
story-telling qualities of scientific narratives by analogy with
human history neglect the extent to which /Awman history
depends on central subjects whose robustness to alternative
periodization is taken for granted. The scientific revolution
may be problematic as a central subject, but Isaac Newton
is not. Gibbon, in a famous passage on the difficulties of
writing ancient history, shows just how far the human his-
torian may justifiably take the central subjects for granted:

The confusion of the times, and the scarcity of authen-
tic memorials, oppose equal difficulties to the histo-
rian, who attempts to preserve a clear and unbroken
thread of narration. Surrounded with imperfect frag-
ments, always concise, often obscure, and sometimes
contradictory, he is reduced to collect, to compare, and
to conjecture: and though he ought never to place his
conjectures in the rank of facts, yet the knowledge of
human nature, and of the sure operation of its fierce
and unrestrained passions, might, on some occasions

JAMES R. GRIESEMER

supply the want of historical materials (quoted in Ki-
ester, 1980:331; emphasis added).

In this passage, knowledge of human nature and the op-
eration of its passions stand as knowledge of mechanisms
which drive human history. It 1s precisely because this knowl-
edge is available for persons that Gibbon has confidence in
his central subjects. When human history departs from per-
sons and their biographies as central subjects, however, the
theoretical issues are just as profound in human history as
in other historical sciences (Hull, 1975), Indeed, it is pre-
cisely the departure from familiar and accepted central sub-
jects that leads historical scientists to models which reduce
the complexity of mechanisms to a manageable level. The
first step in doing this, as | argued above, is to choose periods
within which the modes of mechanism are relatively homo-
geneous.

In one of the most famous attacks on pure story-telling
and defenses of the relationship between narrative and de-
velopmental mechanics, Gould and Lewontin (1979) argue
that alternative “atomizations” of organisms into traits sup-
port different adaptive evolutionary scenarios (or even non-
adaptive ones). Their favorite example of a non-character,
the human chin, tells the story. An idealized and simplified
model of development is tacitly assumed in which any adult
trait describable by a morphologist is seen as emerging by
a simple unfolding. This model takes the periods of trait
development to be whole-organism life cycles. The crude
mechanics of trait development in this example amounts to
little more than the description of chin development as a
gradual, continuous unfolding or growth over the whole-ani-
mal life cycle (with selection typically acting only on the
fully developed adult morphology). From such descriptions,
an evolutionary narrative of chin evolution could be con-
structed. The stages of the narrative are whole-organism life
cycles. The genetic mutations that alter the adult chin mor-
phology of offSpring are the mechanical events within stages
relevant to the evolutionary process. There is no localization
of mutation events within a life cycle because life cycles on
this model do not have distinguishable periods within them,
although the mechanics of mutation suggest that pre- and
post-mutation sub-periods could be identified.

Gould and Lewontin observe that “the chin” develops as
the interaction product of two growth fields, the alveolar and
dentary. These may be under complex genetic control that
has a temporal structure within a life cycle. If true, this struc-
ture is not well-represented by the idealized model described
above. They apply this observation in an argument against
Panglossian adaptationism: if development says “no trait
there,” then it is folly for evolutionists to construct a narrative
interpretation (.¢., adaptation explanation) of its evolution.
Of course, it also does not follow from developmental me-
chanics alone that there /s a trait there. Gould and Lewontin’s
developmental mechanics does not rule out the chin as an
adult trait subject to selection any more than the traditional
mechanics rules it in.!> The critical choice, as John Damuth
points out, is in which developmental mechanics implies
something important about how natural selection actually has

PERIODIZATION AND MODELS
acted. “For if one knows,” Damuth writes,

from careful natural history or manipulative or func-
tional studies how and to what degree selection is op-
erating on a character, one knows both the proper scale
of periodization and the character’s status as a (con-
temporary, at least) adaptation. How to infer past his-
tory of selection is a difficult issue— but no one
should be able to get away with either “Just So” nar-
ratives or “Just So” periodizations. To do so would
be to claim a priori truth for one’s theoretical models
without evidence. (Damuth, pers. comm.)

We can also draw a wider consequence from the chin ex-
ample for the nature of historical explanation than do Gould
and Lewontin. If they are right in their assertion that there
has been no selection on the chin as an adult trait, then the
whole-organism life cycle is too coarse-grained a periodiza-
tion to identify some events relevant to the evolutionary nar-
rative of the hominid lineage. The chin is ex /ypothesi two
interacting traits in development, not one which simply
emerges as an unfolding. Periodization into whole-organism
life cycles (generations) leads to the bad choice of simple
unfolding as the developmental “mechanics” of chin forma-
tion. A periodization that results in a narrative better suited
to the Gould-Lewontin hypothesis 1s one that takes within-
organism developmental stages as the relevant units for the
evolutionary narrative. That is, organismal generations are
not appropriate temporal units for describing chin evolution
because whole life cycles are not temporally homogeneous
with respect to kinds of mechanical events (mutations, in-
teractions of cells of the two fields) that cause the adult chin
state.

Once organisms are themselves divided up into develop-
mental stages, the effects of mutations on different develop-
mental stages can be included in a mechanically articulated
description of the production of an adult chin. But then the
evolutionary narrative will change too, because it will have
the resources to refer to events within life cycles as evolu-
tionarily significant, that is, as bearing significance for
changes in the lineage of human organisms leading to the
current evolutionary state of the chin. This interaction be-
tween developmental model and evolutionary time periodi-
zation is one window on understanding the nature of evolu-
tion as a historical science.

4. Homogeneous Mechanism in Evolution and
Development

The theoretical framework of historical narrative is perio-
dization, the marking of time into the units used in the nar-
rative. The logical structure of narrative sentences involves
relating events in different time units of a periodization. |
suggested that part of the function of periodization is to di-
vide time into stages that are mechanically “homogeneous,”
that is, stages within which a single “ahistorical” kind of
mechanistic description can be produced.'°

In this section | want to illustrate with examples from de-

to
n

velopmental biology how the choice of periodization is de-
signed to facilitate mechanical description and explanation
within periods. In most cases, the role of periodization is
obscured by the fact that temporal stages are initially iden-
tified in morphological or spatial terms, and only sub-
sequently measured in clock units. Moreover, there is often
a complex interplay between the choice of a periodization
and commitment to a particular mode of mechanical descrip-
tion, as is illustrated by the chin example described above.
Possession of a convenient tool for mechanical analysis may
elicit consideration of alternative periodizations. In the late
19th century, for example, the use of microscopy and camera
lucida techniques gave rise to arguments that differentiation
occurs earlier in development than does the appearance of
germ-layers. And by the same token, a well-entrenched pe-
riodization can lead to a search for new tools of mechanical
description, for example the use of patterns of homeobox
gene expression to establish evolutionary homologies. Let
us now consider these two illustrative examples in more de-
tail.

The first illustration mentioned comes from the history of
embryology at the turn of the century. In the 1880s and *90s,
C.O. Whitman, E. B. Wilson, T. Boveri, and others devel-
oped an approach to early embryology that Whitman termed
“cell lineage studies” (Maienschein, 1978, 1991). Whitman
doubted Haeckel’s idea that the embryo was homogeneous
prior to visible germ-layer differentiation and began to study
the earliest cell divisions to get clues about how the germ-
layers formed. Each cell division was studied microscopically
in a number of invertebrates and the pattern of emergence
of germ-layers was correlated with cleavage patterns, move-
ments of cytoplasm, and so forth. But as cell division pro-
ceeds, the complexity of the descriptive task overwhelms
observation. Cell-lineage work shifts from tracking each cell
division to tracking layers at some point in development,
perhaps coincident with the visualization of the germ-layers.
The net result was that the cell lineage workers extended
the time of differentiation in the embryo to much earlier than
Haeckel had, and linked it to the cell divisions that could
be observed individually rather than to the layers that could
be observationally distinguished. In short, the periodization
of development had been changed to include a period of
differentiation prior to germ-layer differentiation and the me-
chanics had shifted to a lower level, from the histological
level of germ-layers to the cellular level of cleavage divi-
sions. Instead of a relatively few temporal divisions in the
germ-layer-based periodization (e.g., appearance of distinct
endoderm and ectoderm as one “tick mark” and then emer-
gence of a distinct mesoderm as another), the cell-lineage
work identified numerous temporal divisions (in the limit,
one for each cell generation). It also pushed and compressed
the stages of germ-layer differentiation into a much earlier
period of development.

Part of the impact of this work on evolution resulted from
the fact that at the cellular level, germ-cells can be considered
a germ-layer and the differentiation of germ-cells became a
joint problem of evolution and development at that level.
Treatment of germ-cells as a “layer” became a prominent

part of the theories of Boveri and Weismann in their inter-
pretations of the role of development in the segregation of
the germ-line and somatic-lines. But the state space in which
these theories are significant depends on the periodization
of cell-lineage work. No matter how much a part of “de-
scriptive embryology” it appears to be, this work has a deeply
theoretical impact through its substitution of a new periodi-
zation.

Consideration of alternative developmental perspectives
and mechanics of the sort described by the cell-lineage work-
ers leads to different evolutionary narratives. Leo Buss’s
(1987) book, The Evolution of Individuality, is an example
of how a re-periodization of development can lead to a novel
evolutionary narrative. By surveying the diversity of times
of segregation of germ-line and soma in a variety of taxa,
Buss was able to use a “Weismannian” periodization of de-
velopment to produce an evolutionary narrative that includes
a previously unacknowledged level of selection. In Buss’s
alternative periodization, the units upon which natural selec-
tion can operate change between developmental periods and
an important effect of the model is to provide the framework
for a unified mechanical account of the operation of selection
within developmental periods — before vs. after germ-line
segregation — that is more fine-grained than the standard se-
lectionist framework.

The tacit periodization of development in the typically ide-
alized, population geneticist’s evolutionary theory is that first
there is the zygote, then development happens (by cell
growth, division and differentiation), and then natural selec-
tion operates among variant (adult) phenotypes, to yield a
differential propagation of genes into the next generation.
(To find examples of this periodization, look at the life cycle
diagrams in textbooks of population genetics and evolution.)
Selection under this model is inevitably at the organism level
because there is no distinction in the tacit model of devel-
opment that would suggest the existence of entities capable
of serving as units of selection (below the level of organisms,
but above that of individual cells or genes). However, it 1s
well known that selection can affect any stage of develop-
ment and that pattern can be laid down even betore fertil-
zation. Buss’s recognition of developmental diversity in the
timing of germ-line separation, of the conservatism of early
ontogeny and the diversity of late ontogeny, leads to vari-
ability in developmental chronology and to an alternative
The timing of
germ-line separation relative to other processes of differen-
tiation marks the transition from one developmental stage to
another in Buss’s periodization of development that supports
the new evolutionary narrative

Between the evolutionary emergence of cells from the pri-
mordial soup and the emergence of metazoa from colonies
of cells, there must have been a stage in which cell lines
competed for germ-line status. If development ts divided into
early ontogeny (prior to germ-line/somatic-line separation)
and late ontogeny (after separation), then there ts the possi-
bility of a new unit of selection: the cell-lineage within a
population of such lineages (colonial proto-organism). The
variability in present developmental mechanisms reflects the

periodization of the evolutionary process

JAMES R. GRIESEMER

historically diverse outcomes of the process of natural se-
lection operating within that stage of development. More im-
portantly for understanding narrative, the significance of his-
torical events in which particular cell lines succeeded in
capturing germ-line status is established in terms of the cur-
rent state of multi-cellular organisms at a new, emergent level
of evolutionary organization. Level of organization is thus
intimately tied to the structure of development, which is re-
vealed by formulating an alternative periodization. The al-
ternative periodization suggests a new mechanical analysis
of the behavior of cell lines toward one another and a new
interpretation of the multi-level process of selection. Thus
our understanding of the evolution of animal individuals must
be given an altered historical narrative, one that takes into
account the new periodization and includes reference to the
new stage 1n its interpretation of the sequence of evolutionary
events.

Turning now to our second illustrative example, one way
to link the articulation of development and evolution to sys-
tematics, through explicit analysis of periodization, 1s shown
inacommentary by Slack, Holland and Graham (1993). They
argue that recent work on the genetics of positional infor-
mation coded by the Hox gene cluster in developing animal
embryos reveals a shared derived character common to all
animals. They propose that,

/an animal is an organism that displays a particular
spatial pattern of gene expression, and we define this
pattern as the zootype. The zootype is expressed most
clearly at a particular stage of embryonic develop-
ment: the phylotypic stage (1993:490).

It is a remarkable fact about the Hox genes that there ts
a 1:11 relation between their spatial ordering in the genome
(from 3’ to 5’ ends of the cluster in the chromosome), the
relative timing of their expression, and the anterior limit of
their expression domains along the anterior-posterior axis of
the embryo. Using these genes, molecular systematists are
able to discover and analyze novel homologies among taxa
so distantly related that traditional morphological assessment
is difficult if not impossible (Holland, 1996:63—70; see also
Holland, 1992; Tabin and Laufer, 1993).

I do not wish to evaluate here the evidence for the Slack,
Holland and Graham zootype thesis, or review the difficulties
with their morphological criteria for the phylotypic stage (a
project the authors trace to Etienne Geoffroy St. Hilaire), or
evaluate their claimed synapomorphy. Holland (1996:63—70)
suggests that the scope of successful taxonomic analyses of
homology probably lies somewhere between Geoffroy’s wide
scope and Cuvier’s much narrower scope, but perhaps the
truth lies much closer to Geoffroy than systematists once
thought. There are two points relevant in the present context.
First, Slack et al. are refining the developmental periodization
relevant to evolutionary narratives beyond Buss’s periodiza-
tion discussed above. In addition to Buss’s pre- vs. post-
germ-line separation periods, which we saw are relevant to
the construction of narratives describing the evolution of ant-
mal individuality, Slack et al. add the phylotypic stage, in

PERIODIZATION AND MODELS

which peak expression of the zootypic Hox genes occurs, as
relevant to the phylogenetic differentiation of animals. This
stage will typically occur at a different point in time from
the separation of germ and soma, so adding it to Buss’s
periodization is a refinement. Moreover, in general, the dis-
covery of further genetic events of the same sort that Slack
et al. identify pushes developmental periodization to a mo-
lecular level which is likely to yield further refinements. In
particular, the phylotypic stage is a period defined in terms
of the mechanical analysis of the expression of a single class
of genes. The Slack et al. periodization is in fact built on
the foundation of a unitary mechanical analysis of gene ex-
pression. One can imagine other stages of development de-
fined by expression of other classes of genes, as indeed recent
works on the molecular biology of development acknowledge
(Lawrence, 1992).

Second, Slack et al. tie their zootype thesis explicitly to
the issue of phylogeny reconstruction by arguing that the
zootype is a shared derived character (synapomorphy) for
the kingdom Animalia. From the point of view of cladistic
systematics, synapomorphies provide the best evidence of
phylogenetic relationship. From the point of view of com-
parative developmental biology, synapomorphies provide the
best opportunity for a common mechanical analysis and thus
for an integrating narrative. Discovery of synapomorphies in
development is facilitated by judicious choice of periodiza-
tion. It is probably not accidental that the evolutionists most
concerned with the relationship between evolution and de-
velopment at the present time are systematists who make
use of cladistic methods. Whether the zootype hypothesis
is true or not, the Slack et al. claim illustrates how devel-
opmental periodization, the postulation of a new state space
in which to frame theoretical models that play a role in his-
torical narratives, 1s implicated in phylogenetic hypotheses.
It is plausible to infer that phylogenetic arguments make
some tacit assumption about developmental periodization,
just as Sober (1988) argued that all cladistic analyses must
assume a model of character evolution.

Several features of the Slack et al. analysis bear on the
role of periodization in setting the terms for mechanical ex-
planations within stages and evolutionary narratives among
stages. I do not want to claim on their behalf, however, that
their intent is to revise the developmental periodization in
order to discover novel evolutionary mechanisms, although
I think it is likely that periodizations may often produce that
heuristic effect. Rather, my aim is to illustrate that choice
of periodization is tied to the choice of a particular devel-
opmental mechanics (Hox gene expression) that is used as
a tool for phylogeny reconstruction. Slack et al. point out
that there are a number of morphological definitions of
Sander’s concept of the “phylotypic” stage:

... the stage of development at which all major body
parts are represented in their final positions as undif-
ferentiated cell condensations, or the stage after the
completion of the principal morphogenetic tissue
movements, or the stage at which all members of the

27,

phylum show the maximum degree of similarity.
(Slack et al., 1993:491)

Each of these definitions identifies a form of mechanical
description (cell condensation states, morphogenetic tissue
movements, degree of similarity) that can be uniformly ap-
plied within a stage. By this | mean that each mode of me-
chanical description requires a small set of procedures and
methods that can be applied to embryological materials
throughout the stage, e.g., a single set of histological-micro-
scopical techniques for visualizing condensation states. This
much is required for any effective mechanical science. But
part of the purpose in identifying the phylotypic stage is to
facilitate narrative as well as mechanical explanation. The
mechanical descriptions are applied to identify the phylotypic
stage in different phyla and then used to make comparisons
among the phylotypic stages of various phyla to construct a
narrative history of the Animalia. The mechanical descrip-
tions lead to identifying the following classically described
developmental stages as phylotypic stages:

... the tailbud stage for the vertebrates; the fully seg-
mented germband stage for insects; the fully seg-
mented, ventrally closed stage for leeches; or the
nematode after the completion of most embryonic cell
divisions. (ibid.)

It is not essential that the same mode of mechanical de-
scription be applied to material from different phyla, but only
that there is some such mode of description for each. How-
ever, the observation critical for the construction of the evo-
lutionary narrative (presented in the form of a hypothetical
cladogram) is the correspondence between the embryonic
stage at which the zootype is most clearly displayed and the
phylotypic stage. Slack et al. write:

The genes of the zootype are not, in general, activated
in the earliest stages of development, and although ex-
pression may persist for some considerable time, the
peak expression, and the simplest pattern of expres-
sion, is displayed at the phylotypic stage. This asso-
ciation gives us confidence that the independent pro-
posals of evolutionary conservatism of the zootype and
of the phylotypic stages are indeed well founded.

(ibid.)

Since the zootype is a pattern of gene expression, the cor-
respondence between zootype and phylotypic stage offers
the possibility of a single mode of mechanical description
for within-phylotypic stage analysis that can then be used to
construct an evolutionary narrative based on analysis of dif-
ferences in the Hox gene clusters of different phyla. The
stages of this evolutionary narrative are the durations of taxa
exhibiting a particular genetic structure. The mechanical
events that are linked in the narrative are changes in genetic
structure: from the origin of helix-turn-helix genes predating
the prokaryotes, to the origin of homeobox genes predating

the fungi and green plants, to the origin of Hox cluster genes
and the zootype predating the branching of Cnidaria, Platy-
helminthes, and higher metazoa (Slack et al., 1993:492, fig.
4).

5. Conclusion

I have argued that joint consideration of evolutionary nar-
ratives and developmental mechanics illuminates the nature
of historical science. The crux lies in the coordinating role
of periodizations in formulating both theoretical models and
narratives to describe the integration of historical individuals
persisting across periods with mechanical accounts of proc-
esses of change occurring within periods. If, as Laudan
(1992) has argued, the question at issue in understanding
historical science is how knowledge claims in historical se1-
ences are warranted, then I agree that there is no particular
significance to the distinction between historical and ahisto-
rical sciences. But the aim of the distinction that motivates
my interest 1s quite different.

My concern has been to introduce the idea that the his-
torical sciences are deeply theoretical, if one interprets pe-
riodization as setting the framework of state space type in
which theoretical models and narratives can be expressed.
The models serve, among other functions, to identify events
that can be explained mechanically and also the central sub-
jects of narratives. The models mediate between data and
theory and the periodization coordinates the mediating rela-
tions. Rather than seek protection from the critics of historical
science by emphasizing the distinctiveness of historical phe-
nomena or data, | propose to base a defense of the historical
sciences on the distinctive characteristics of their theoretical
models. What is distinctive about historical science is that
its periodization models coordinate narrative and mechanistic
accounts and that historical scientists accept the narrative
component as a contribution to understanding.

Whether any defense of the historical sciences is effective
can only be decided by trying it out. Since what makes a
science historical is a pragmatic commitment to narrative
understanding, it will always be open to rejection by some
serentists and acceptance by others. Historicity entails a com-
mitment to a mode of understanding, not a mode of inference
ora theory of reality. It will therefore prove a more attractive
strategem, in: my view, to defeat the argument that historical
selence 1s “bad science” by advertising the virtues of coor-
dination of narration and mechanism by periodization and
models than to seek ontological or epistemological priority
or distinctiveness for the historical sciences.

Notes

|. My starting points on the notion of narrative are: Danto,
1968/1985 and White, 1987. See White, 1987 for a review
of historiographic positions opposed to narrative.

2. On the former see, for example, Hull, 1975, Gould et
al, 1987, O° Hara, 1988, Gould, 1989, Ereschefsky, 1992.
On the latter see Hempel, 1942, Hull, 1992, Richards, 1992.

JAMES R. GRIESEMER

3. See Griesemer, 1990 for a similar demonstration of theo-
retical activity in apparently untheoretical natural history.

4. This term was introduced by Lloyd (1988) in analogy
with van Fraassen’s concept of model type, and adopted by
van Fraassen (1989) to indicate that many clusters of models
which specify a theory can have the same state space, e.g.,
population genetics models expressed in a gene frequency
state space. Periodization is an important step in the delimi-
tation of a state space type.

5. See White, 1987, ch. 2 on the distinction between annal,
chronicle, and narrative.

6. Rudwick, 1985, Secord, 1986, and Oldroyd, 1900 dis-
cuss these episodes, each focusing on a different one.

7. For an example of such a description, see Balinsky,
[9756p e171.

8. Minelli, 1996:55—61.

9. Ibid.

10. These concern what things can exist at a point in state
space, e.g., What combinations of gene frequencies are pos-
sible.

11. These concern which sequences of states are possible
in a given state space, e.g., if motion is continuous and po-
sition is defined along a real-valued axis, then to get from
A to B, an object must pass through all the intermediate
points between them. In population genetics, gene frequen-
cles range from 0 to | continuously, but they can jump from
generation to generation, because populations are finite (so
a mutation causes a discrete change in the gene frequency).

12. Minelli, 1996:55—61.

13. Historians and philosophers of science may reflect on
Kuhn’s revolution model of science as a case study of this
process. Many of Kuhn’s critics reyected his overall account
because of objections to his psychological account of the
revolutionary transition, while accepting (more or less) his
description of “normal science.” The Gestalt model was a
sort of place-holder for a serious attempt to account for dra-
matic and creative change in science. Kuhn’s achievement
was to string together normal science, crisis, and revolution
into a coherent narrative structure for many scientific epi-
sodes. Many who “reject” the model misunderstand the heu-
ristic value of making such a periodization of scientific
change.

I4. This pair of relations linking phenomena or data,
model, and “high” theory is explored to great depth by Nancy
Cartwright (1983). The analogy here rests largely on the fact
that periodization sets the state space framework for models.
Empirical substructures are the observable elements of theo-
retical models.

15. | wish to thank John Damuth (pers. comm., 5 Aug.
1994) for helpful discussion on this point.

16. “Ahistorical” is in quote marks because on some (rela-
tively trivial) interpretations of historicality, every mecha-
nistic science 1s historical. See Griesemer, 1996 for discus-
sion and references.

PERIODIZATION AND MODELS
Acknowledgments

I thank John Damuth, Elihu Gerson, Michael Ghiselin,
Nick Holland, Bob O’Hara, Alessandro Minelli, Dave Wake,
and especially an anonymous reviewer for helpful discussion
and comments on the manuscript and Michael Ghiselin,
Giovanni Pinna, and Marvalee Wake for organizing the
workshop. The manuscript was written in 1992—93 during a
fellowship at the Wissenschaftskolleg zu Berlin and revised,
in part, while [ was a fellow of Collegium Budapest in 1994—
95. I wish to thank the Rektors and fellows of both institutes
for providing such splendid working conditions.

Literature Cited

Balinsky, B. I. 1975. An Introduction to Embryology, 4th Edition.
W.B. Saunders, Philadelphia.

Benjamin, W. 1968. Illuminations. Harcourt, Brace & World, New
York.

Buss, L. W. 1987. The Evolution of Individuality. Princeton Uni-
versity Press., Princeton.

Cartwright, N. 1983. How the Laws of Physics Lie. Oxford Uni-
versity Press, New York.

Danto, A. C. 1985. Narration and Knowledge (Including the Integral
Text of Analytical Philosophy of History), 2nd Edition. Columbia
University Press, New York.

Dobbs, B. J. T. 1994. Newton as final cause and first mover. Isis
85:633-643.

Ereshefsky, M. 1992. The historical nature of evolutionary theory.
Pages 81—99 in M. H. Nitecki and D. V. Nitecki, editors. History
and Evolution. State University of New York Press, Albany.

Gerson, E. M. MS. The American System of Research: Evolutionary
Biology 1890-1950.

Ghiselin, M. T. 1969. The Triumph of the Darwinian Method, Ist
Edition. University of California Press, Berkeley.

Giere, R. 1988. Explaining Science: a Cognitive Approach. Uni-
versity of Chicago Press, Chicago.

Gould, S.J. 1977. Ontogeny and Phylogeny. Harvard University
Press, Cambridge.

———.. 1989. Wonderful Life: the Burgess Shale and the Nature
of History. Norton, New York.

Gould, S. J., N. L. Gilinsky, and R. Z. German. 1987. Asymmetry
of lineages and the direction of evolutionary time. Science
236:1437-1441.

Gould, S. J. and R. C. Lewontin. 1979. The spandrels of San Marco
and the Panglossian paradigm: a critique of the adaptationist pro-
gramme. Proceedings of the Royal Society of London 205:581
598.

Griesemer, J. R. 1990. Modelling in the museum: on the role of
remnant models in the work of Joseph Grinnell. Biology and
Philosophy 5:3—26.

— . 1996. Some concepts of historical science. Memorie della
Societa Italiana de Scienze Naturali e del Museo Civico di storia
Naturale di Milano 27:60-69.

Hempel, C. G. 1942. The function of general laws in history. The
Journal of Philosophy 39:35-48.

Holland, N. D. 1996. Homology, homeobox genes, and the early
evolution of the vertebrates. Pages 63-70 in M. T. Ghiselin and
G. Pinna, editors. New Perspectives on the History of Life: Essays
on Systematic Biology as Historical Narrative. California Acad-
emy of Sciences, San Francisco.

Holland, P. W. H. 1992. Homeobox genes in vertebrate evolution.
BioEssays 14:267-273.

Hull, D. L. 1975. Central subjects and historical naratives. History
and Theory 14:253-274.

———. 1992. The particular-circumstance model of scientific ex-
planation. Pages 69-80 in M.H. Nitecki and D. V. Nitecki, edi-
tors. History and Evolution. State University of New York Press,
Albany.

Kiester, A. R. 1980. Natural kinds, natural history and ecology.
Synthese 43:331-342.

Latour, B. 1993. We Have Never Been Modern. Harvard University
Press, Cambridge.

Laudan, R. 1992. What’s so special about the past? Pages 55-67
in M.H. Nitecki and D. V. Nitecki, editors. History and Evolu-
tion. State University of New York Press, Albany.

Lawrence, P. A. 1992. The Making of a Fly. Blackwell, Oxford.

Lloyd, E. A. 1988. The Structure and Confirmation of Evolutionary
Theory. Greenwood Press, New York.

Maienschein, J. 1978. Cell lineage, ancestral reminiscence, and the
biogenetic law. Journal of the History of Biology 11:129—158.

———. 1991. Transforming Traditions in American Biology,
1880-1915. The Johns Hopkins University Press, Baltimore.

Mayr, E.
1506.

1961. Cause and effect in biology. Science 134:150]—

Minelli, A. 1996. Segments, body regions, and the control of de-
velopment through time. Pages 55-6] in M. T. Ghiselin and G.
Pinna, editors. New Perspectives on the History of Life: Essays
on Systematic Biology as Historical Narrative. California Acad-
emy of Sciences, San Francisco.

O’Hara, R. J. 1988. Homage to Clio, or, toward an historical phi-
losophy for evolutionary biology. Systematic Zoology 37:142-
[55:

Oldroyd, D.R. 1990. The Highlands Controversy: Constructing
Geological Knowledge through Fieldwork in Nineteenth-Century
Britain. University of Chicago Press, Chicago.

Richards, R. J. 1992. The structure of narrative explanation in his-
tory and biology. Pages 19-53 in M. H. Nitecki and D. V. Nitecki,
editors. History and Evolution. State University of New York
Press, Albany.

Rudwick, M.J.S. 1985. The Great Devonian Controversy. The
Shaping of Scientific Knowledge among Gentlemanly Specialists.
University of Chicago Press, Chicago.

Secord, J. A. 1986. Controversy in Victorian Geology: the Cam-
brian-Silurian Dispute. Princeton University Press, Princeton,

Slack, J.M. W., P. W.H. Holland, and C. F. Graham. 1993. The
zootype and the phylotypic stage. Nature 361:490-492.

Sober, E. 1988. Reconstructing the Past: Parsimony, Evolution, and
Inference. MIT Press, Cambridge.

JAMES R. GRIESEMER

Tabin, C. and E. Laufer. 1993. Hox genes and serial homology.
Nature 361:692-693.

Van Fraassen, B. C. 1989. Laws and Symmetry. Clarendon Press,
Inventory.

Vermeij, G. J. 1987. Evolution and Escalation: an Ecological His-
tory of Life. Princeton University Press, Princeton.

White, H. 1987. The Content of the Form: Narrative Discourse and
Historical Representation. Johns Hopkins University Press, Bal-
timore.

Wimsatt, W. C. 1981. Robustness, reliability and overdetermination.
Pages 124-163 in R. Brewer and B. Collins, editors. Scientific
Inquiry and the Social Sciences. Josey-Bass, San Francisco.

—. 1987. False models as means to truer theories. Pages 23-55
in M. Nitecki and A. Hoffman, editors. Neutral Models in Biol-
ogy. Oxford University Press, London.

THE OLDEST FOSSIL ANIMALS
IN ECOLOGICAL PERSPECTIVE

Mikhail A. Fedonkin

Paleontological Institute
Russia Academy of Sciences
Profsoyuznaya ulitza 123
Moscow 117647 Russia

Introduction

The ecological aspect of the history of life attracts growing
attention as a missing piece in a puzzle of pattern and process
in evolution (Eldredge, 1989). The whole fossil record dem-
onstrates a fundamentally interactive relationship between
the biota and the abiotic environment, and the primary role
of organisms in changing the earth from the strange and
unfamiliar planet that it was in the Precambrian to what we
see today. We are beginning to understand the feedback
mechanisms of the global environmental control exerted by
the biota.

Paleoecology, defined as the paleontology of the biosphere
as a whole, allows us to study global processes that cannot
be revealed by such ecological approaches as monitoring that
focus upon the present. Long term cycles, resistant trends,
and events of global importance in the biosphere are the
privilege of paleoecology.

However, traveling backward in time we encounter biotas
less and less similar to the Recent one. This is why the role
of actualistic interpretation of fossils in paleolecological re-
constructions should be decreasingly emphasized as we go
backward in time, and increasingly complemented by infor-
mation from sources other than the fossil record. The mul-
tidisciplinary approach that is the core of the Recent paleoe-
cological methodology is especially legitimate with respect
to the oldest parts of the history of the biosphere (Schopf
and Klein, 1992).

The history of the organic world can be represented by
the U-pattern and the N-pattern for the procaryotic and eu-
caryotic worlds, respectively (Fig. 1).

The U-pattern of the procaryotic world is characterized by
rapid (geologically instantaneous) radiation of the pro-
caryotes and their long history of persistence without essen-
tial changes throughout most of geological time. This is par-
ticularly true of the Cyanobacteria, which seem to have
stayed the same both morphologically and ecologically for
more than 3.5 billion years.

The N-pattern evolution of the eucaryotes reflects two
processes, namely the decline in diversity of the relict groups
(the left half of the letter N) and the growing diversity of
the new groups (the right half of the N). This pattern means
that the general biotic diversity at any time level includes
both relict groups and newborn taxa. This seems to be the
case even for the earliest faunas, including the oldest known
metazoans of the Vendian period (Fedonkin, 1992). There-
New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

fore it helps to distinguish those Vendian taxa which gave
rise to the Paleozoic groups on the one hand, from those
that represent remnants of Pre-Vendian periods of metazoan
evolution on the other. The N-pattern approach can be ap-
plied to the whole hierarchy of the eucaryotes at every level
from the species upward. For instance, Cnidaria and Cteno-
phora, the two phyla representing the diploblastic grade of
organization can be considered relics of the first Eumetazoa
(Fedonkin, 1985; Rieger, 1994).

The closer to the present the smaller is the proportion of
the relict groups in the animal world. Conversely, when we
move back in time we encounter more and more fossil meta-
zoans with unusual body plans. Does this mean that we are
encountering forms with unusual physiologies which have
no equivalent relationship among Recent animals? In fact,
there should be some positive correlation between novelties
in body plan and innovations in physiology, so far as these
are reflected in the differentiation of the Recent metazoans.
Thus we may confidently predict that we will encounter dif-
ferent physiologies among the oldest metazoans. This in turn
implies a different relationship of the earlier metazoans to
the paleoenvironment. This circumstance necessitates that we
reconstruct the abiotic and biotic factors of the global and
local paleoenvironments of the Precambrian if we are to un-
derstand the nature and early history of the oldest metazoans.

The origin of animals: change in priority of the
questions “when, Be

99 6

where,” and “who?

The origin of the metazoans is an old but unsolved prob-
lem, in spite of substantial progress in modern paleontologi-
cal and neontological sciences alike. For generations of bi-
ologists the major question related to the problem has been:
“who” was at the very base of the metazoan tree — what
was the grade of organization of the first multicellular ani-
mal? In the nineteenth century the priority of this “who”
question over alternatives (when and where) was connected
with the leading role of comparative anatomy and embryol-
ogy in the study of the problem of the origin of the Metazoa.
Traditionally these sciences were oriented toward the mor-
phology and morphogenesis of Recent animals, rather than
toward the biohistorical or ecological aspect of the problem.
An additional factor that favored the priority of the “who”
question was the dominating paradigm of the monophyletic
origin of the Metazoa: the notion of a common ancestor.

Since the end of the nineteenth century numerous phylo-

ws)
to

Time

PROCARIOTES

Diversity

Time

EUCARIOTES

recent
groups

MIKHAIL A. FEDONKIN

B
relict
groups
Diversity
Time
METAZOANS
Triploblastica
Cc
Diploblastica
Diversity
FIGURE 1. U- and N- patterns of the diversity through the history of the organic world. A. U-pattern of the procaryote history. B. N-pattern
of eucaryote history. ¢

N-pattern of the metazoan history

OLDEST FOSSIL ANIMALS

genetic trees for the animals have been constructed, differing
both in their trunks and their crowns. The new science of
histology made some contributions to the growing diversity
of this orchard, but did not give decisive arguments which
would allow us to separate the “good” trees from the “bad”
ones. The rise of molecular biology has provided a new in-
strument for the quantitative measurement of the degree of
phylogenetic proximity. The molecular clock approach
seemed to promise a way to establish the time of major
branchings in the phylogenetic tree. Unfortunately, along
with the general methodological difficulties related to prob-
lems of homology and analogy that are common for most
comparative methods, two fundamental assumptions con-
cerning molecular clocks remain uncertain: the constant rate
of the clocks, and their dependence upon generation time.
In addition, we have no idea yet of how the whole phylo-
genetic picture could be influenced by various symbiogene-
ses which seem to have been widespread in the Precambrian.

This situation suggests that the problem of metazoan or1-
gins and the monophyletic paradigm and approaches men-
tioned above should be set aside. Instead we should be ad-
dressing the problems of the time of origin of the
newly-appeared metazoans, the primary characteristics of
their biotope, their place in the pre-existing ecosystem and
their physiologies and modes of life. All these questions,
including that of the time of metazoan origin, are connected
to the ecological aspect of the problem. The Late Proterozoic
was a period of dramatic environmental change, with strong
differences between the beginning and the end of the con.
So we have to construct essentially different models for meta-
zoan origins for the moments of 1,000 million years and 700
million years ago, as far as we understand the Precambrian
environments. In order to understand how and why it took
place we have to know where it took place — in both space
and time.

PALEONTOLOGICAL DATA

The first traces of life are as much as 3.5—3.8 billion years
old and multicellular fossils first appear in deposits about
2.0 billion years old. However, undoubted trace and body
fossils of metazoans appear at the very end of the Precam-
brian fossil record during the Vendian Period about 620-550
million years ago. The famous Ediacara fauna and its world-
wide equivalents in other Vendian deposits manifest an epi-
sode in the global expansion of the oldest known inverte-
brates. These faunas seem to reflect the stage of evolutionary
stasis which followed the earlier radiation (or series of ra-
diations and extinctions) of older multicellular animals. To
avoid misunderstanding of terms | use the word “Ediacara”
to specify the type of fauna and the term “Vendian” for the
Terminal Proterozoic Period (Sokolov and Fedonkin, 1984).

The Pre-Vendian history of the metazoans still remains
obscure. The search for the oldest animal fossils has mainly
revealed a large number of Proterozoic pseudofossils and
dubiofossils (see critical analyses of the problem in Cloud,
1968; Hantzschel, 1975; Glaessner, 1979a; Fedonkin and
Runnegar, 1992). Nevertheless the morphological characters

of some problematic body fossils imply that animal (and
even metazoan) life existed long before the beginning of the
Vendian Period (Sun, 1986; Sun, Wang, and Zhou, 1986;
Fedonkin, Yochelson, and Horodsky, 1994).

The late Proterozoic decline of the stromatolites, which
was especially dramatic after 0.85 billion years ago, might
be considered indirect evidence of increasing metazoan ac-
tivity (Awramik, 1971; Monty, 1974; Walter and Heys,
1985). This conclusion is open to question because perhaps
some alternative or additional biotic and abiotic environ-
mental factors could have led to the decline of the stroma-
tolite communities as globally dominant in shallow water
ecosystems (Semikhatov and Raaben, 1994; Fedonkin, 1994,
1996).

Extrapolation from the evolutionary rates of the major
Phanerozoic phyla down to the Precambrian led Durham
(1970) to conclude that the common ancestor of the deu-
terostomes lived between 0.8 and 1.7 billion years ago, which
would move the time of metazoan origin to an even earlier
time period. However, recent molecular data indicate that
plants, diploblastic metazoans, and triploblastic metazoans
were produced by three closely-related groups that radiated
almost contemporaneously (Christen, 1994). Discoveries of
Vendian and Cambrian problematic fossil taxa, which cannot
be placed in the existing system of metazoans, suggest that
we need a more complex scenario of early metazoan evolu-
tion than the one that was invoked twenty to thirty years
ago.

NEONTOLOGICAL DATA

Data from the comparative anatomy, embryology and his-
tology of Recent metazoans allow us to work out phylogenies
with reasonably good resolution within separate major clades.
However, the variety of incompatible models indicates the
inadequacy of classical neontological methods for solving
the problems of the origin of the major phyla and the early
history of the metazoans. Excessive parallelism in somatic
evolution and too few homologous characters that can be
recognized in different phyla make us look for complemen-
tary approaches.

Recently Valentine (1991, 1994) attempted to avoid these
methodological difficulties. His approach was based upon
the idea of Bonner (1965, 1988) that the best single metric
of an organism’s complexity is the number of cell types that
it possesses. Valentine extrapolated from a modern grade of
210 cell types (observed in recent mammals) through the
average of 30 and 50 cell types that should have existed in
the producers of the oldest horizontal trails, and the primitive
higher invertebrates respectively. He obtained an estimate
for the origin of metazoans of 680 million years ago.

MOLECULAR DATA

The last decade has been marked by intensive study of
molecular phylogenies based upon moleculecular sequences,
especially 18S ribosomal RNA (Field et al., 1988) and 28S
rRNA (Christen et al., 1991). These data have been inter-

preted using distance methods (Field et al., 1988; Raff et
al., 1989), cladistic approaches (Ghiselin, 1988, 1989), and
parsimony techniques (Patterson, 1989; Smith, 1989; Lake,
1990; Bergstrém, 1994).

Molecular analysis has all the weaknesses in common with
comparative methods in general; that is why the orchard of
phylogenetic molecular trees is growing even faster than did
that which arose from the classical approaches. Nonetheless
there is some consensus that the metazoans emerged during
a period of intensive diversification, and from the same
branch of the eucaryote stock from which arose vascular
plants and fungi. Evidently the appearance of the tissue grade
of organization was followed by a rapid diversification of
the phyla (Christen, 1994).

Applying molecular clock approaches to cytochrome c and
globin molecules, various authors have estimated that radia-
tion of the living animal phyla took place at least 900—1,000
million years ago, and that the first metazoans could have
appeared even earlier (Dickerson, 1971; McLaughlin and
Dayhoff, 1973: Runnegar, 1982a, 1986). Although these es-
timates have been criticized (Erwin, 1989), the possibility
of a relatively long Precambrian history of the metazoans
still cannot be ruled out (Runnegar, 1992). According to
Chapman (1992), the molecular data suggest an evolution of
eucaryotes much further back in time than was previously
suspected. Therefore we may speculate about the origin of
the eucaryotes around 2.5 billion years ago. and about the
massive radiation of the fungi, photosynthetic protists and
animals between 1.5 and 1.0 billion years ago. Thus meta-
zoans could have appeared no earlier than during that time
interval. Let us now consider the global environmental con-
ditions during the Middle and Late Proterozoic, a period of
time known as the Riphean (1.6—0.65 billion years ago).

Sources of paleoecological information

Paleoecological information can be acquired from rather
heterogeneous sources. The uncertain nature of many Pre-
cambrian fossils compels us to follow non-biological ap-
proaches to paleoecological reconstruction, including
evidence from sedimentology, paleogeography and paleo-
clmatology.

One very important fact is that most of the fossil localities
of the Vendian fauna occur in silicoclastic facies that accu-
mulated in rather cold climatic zones. The only fossil locality
of Vendian metazoans in carbonate rocks 1s thin-laminated
bituminous limestone of the Khorbusuonka Series, Oleniok
Uplift, North of the Siberian Platform (Fedonkin 1985, 1990;
Vodanjuk, 1989), and this locality is less fossiliferous than
those in the silicoclastics. The relatively low diversity of
body fossils and extreme rarity of trace fossils in this series
may indicate that for most of the known Vendian metazoans,
unlike those of the Cambrian, the native biotopes were not
basins with carbonate sedimentation. This unique locality re-
quires further detailed study with respect to paleogeography
and paleoecology, for the basin was dominated by stroma-
tolite communities during most of the post-glacial (Post-

MIKHAIL A. FEDONKIN

Varangerian) portion of the Vendian. The close proximity
of stromatolites, both stratigraphically and paleogeographi-
cally, may indicate conditions unfavorable for metazoans
(see below).

Recent paleogeographic reconstructions (see, for instance,
Torsvik et al., 1992; Torsvik, 1994) demonstrate that during
the late Cambrian and Ordovician a paleo-ocean existed be-
tween Baltica and Siberia. Both plates were situated in the
southern hemisphere and moved northward 5—8 cm per year.
The White Sea region, which has yielded the most diverse
Vendian fauna, was about 55°S and the rest of the Russian
Platform was at even higher latitudes to the south. What is
now northern Siberia (Oleniok Uplift) was oriented toward
the south 490-500 million years ago, and was situated about
30°S. Most of Avalonia, where Precambrian metazoan fossils
are abundant, was more than 60°S. These data mean that
both the Russian Platform and Avalonia, and even the
Oleniok Uplift of northern Siberia, were situated far to the
south, in the colder waters of the Vendian paleo-ocean.

There is strong evidence that the majority of the Vendian
silicoclastic basins that were inhabited by an abundant and
diverse fauna were paleogeographically situated in relatively
high latitudes with a moderately cool climate. The conclusion
that the cold water basins were favorable is supported: 1)
by the presence of purely silicoclastic sedimentation and ab-
sence of carbonates in the various fossiliferous facies; 2) by
the stratigraphic proximity of tillite-bearing deposits below
the beds containing the soft-bodied fauna (Varanger glacial
deposits and their correlatives all over the world) and also,
presumably above beds with Vendian metazoans (glaciation
at the Vendian-Cambrian boundary); and 3) by paleo-
geographic reconstructions based on paleomagnetic data.

More detailed study of the sedimentological control over
the distribution of the Vendian body fossils and trace fossils
indicates that their inhabitants were most abundant in shallow
water environments. Sedimentological analysis of the type
locality of the Ediacara fauna (Golding and Curnow, 1967;
Gehling, 1983, 1988) led to the conclusion that the fossilif-
erous facies represent offshore, storm-wave base environ-
ments and that the organisms were predominantly benthic.
This may be true for the Eidacara assemblage, but it Is not
true for Vendian metazoans as a whole. In sequences else-
where in the world, fossiliferous facies containing the Ven-
dian-type fauna are far more diverse.

The apparent dominance of benthic forms in the Ediacara
assemblage and in its world-wide counterparts may be just
the result of how the fossils happened to be formed, in other
words, a taphonomie artifact. While they were in their living
position, benthic animals deformed the sediment. Next they
got buried alive and undistorted by sediment that was trans-
ported by storms or other events. And then biological deg-
radation proceeded inside the sediment, so that the products
of decay would have been concentrated in the sediment
around the corpse. This led to a faster diagenesis of the sedi-
ment in this particular space, and thus to the fossilization of
the structures Imprinted by the animal that lay buried in the
sediment.

OLDEST FOSSIL ANIMALS
EVIDENCE FROM THE FOSSILS. IN THE VENDIAN
TAPHONOMIC WINDOW

The uncertain nature of many Ediacara forms, and the ta-
phonomic peculiarity that affects the taxonomic work of pa-
leontologists makes us use a taxon-free characterization of
the Vendian metazoan communities. The usefulness of this
approach is demonstrated by the study of both marine and
terrestrial ecosystems (Damuth et al., 1992).

Considering the fossil record in its entirety, the mass pres-
ervation of soft-bodied metazoans in the Vendian deposits
that accumulated in the normal, well-aerated marine basins
was a relatively brief episode in geological history. The
Phanerozoic record does demonstrate a few exceptional cases
of preservation of soft-bodied invertebrates — for instance,
the world famous localities of the Middle Cambrian Burgess
Shale, Pennsylvanian Mazon Creek, Jurassic Solenhoffen,
and some others — but these localities are most unusual,
given the general background of the fossiliferous formations
dominated by skeletal remains. Classical paleozoology 1s pri-
marily the paleontology of skeletons. Therefore what is an
exception in the Phanerozoic part of the fossil record is the
norm for its Precambrian portion, although the taphonomic
window to early animal life seems to have opened for a
relatively short period of time (approximately from 620 to
570 million years). The preservation of abundant soft-bodied
metazoans which is observed in the variety of silicoclastics
appears to have ceased well before the end of the Vendian.
The biotic and abiotic factors which could promote the pres-
ervation of the soft-bodied animals have been discussed in
detail elsewhere (Fedonkin, 1985, 1992; Gehling, 1986,
1991). But in a sense the Vendian taphonomic window was
opened as a result of increasing body size and growing abun-
dance of metazoans. Vast epiplatform basins appeared as the
result of a glacio-eustatic transgression of the sea onto the
continents at the beginning of the Vendian period. Therefore,
both the body size and the total biomass of Precambrian
metazoans could increase very fast during the initial stage
of the pioneering colonization of free ecospace that was made
available. An old idea (since Darwin’s time) that the Pre-
cambrian animals were of small size and thus could not be
preserved, can now be applied to the apparent absence of
Pre-Vendian fossil metazoans — the small size possibly re-
sulting from lower concentration of oxygen in the sea water
(Runnegar, 1982b).

The experience of classical paleontology with Phanerozoic
materials, largely consisting of skeletal remains, makes it
evident that the fossilization of soft-bodied animals is a rare
and exceptional taphonomic event. Given the great variety
of extant non-skeletal metazoan groups which do not have
a fossil record, we would only expect that there would have
been an enormous loss of paleobiological information. Trying
to determine former biotic diversity on the basis of skele-
ton-bearing taxa Is like trying to estimate the population of
ancient Egypt on the basis of the number of pharaohs’ sar-
cophag1.

Mass preservation of soft-bodied metazoans in well-aer-
ated shallow-water sediments distinguishes the Vendian fos-

es)
a

sil record from that of the Phanerozoic. Skeletal remains are
rare in the Vendian. All of these features make it a ta-
phonomically unusual period. The nature of this taphonomic
inversion has been discussed by Wade (1968), Glaessner
(1984), Seilacher (1984), Fedonkin (1985, 1992), and
Gehling (1986, 1991). Major factors responsible for the pres-
ervation of soft-bodied organisms in the Vendian are: 1) ab-
sence or low activity of predators and scavengers (Glaessner,
1984); 2) low levels of biological processing of the sediment
by deposit feeders and other mobile benthos (Fedonkin,
1981); 3) cyanobacterial films which stabilized the sediment
after the burial of the animal (Gehling, 1986) and hindered
the aeration of the bottom, thereby increasing the rate of
hardening of the sediment around the decaying body (Fe-
donkin, 1987, 1994); and 4) absence of active filterers which
could have affected the properties of the water and sedimen-
tological and taphonomic processes similar to those that oc-
curred in the Phanerozoic (Fedonkin, 1985, 1992).

Considering the cold water environment of the Ediacara
fauna we could add a low rate of biological degradation to
this list of biotic factors that promoted the mass preservation
of soft-bodied organisms. Low temperatures are far more
effective inhibitors of decay than anoxia (Kidwell and Bau-
miller, 1990). According to Stanley and Herwig (1994), func-
tional optima of many enzymes participating in bacterial deg-
radation of complex organic compounds are much higher
than the temperatures dominating in Antarctic waters.

Change in these biotic factors at the end of the Vendian
closed the taphonomic window of Ediacara-type preserva-
tion. But what factor or factors opened this window in the
first place? The critical points should be body size and tissue
resistance of the earlier metazoans, which had to be large
and tough enough to leave imprints or to produce preservable
and observable bioturbations of the sediment. On the other
hand, the great preponderance of sedentary forms relative to
pelagic ones in the Vendian faunal assemblages may indicate
a connection between the appearance of the metazoans in
the fossil record and the colonization of the benthic realm
by animals.

Misunderstanding of the tayhonomy of the Vendian meta-
zoans has resulted in misinterpretation of the body fossils
and in erroneous (indeed extravagant) reconstructions of the
anatomy, physiology, and mode of life of these organisms.
Thus, all discoidal fossils were at first interpreted as
medusae, though most of these extremely numerous fossils
were in fact sedentary cup-like polyps or the attachments of
colonial, frond-like organisms such as Charnia, which could
reach 1.3 m in length. Radial cracks in the mesogloea of the
medusoids were interpreted as channels of the gastro-vascu-
lar system. The flat shape of the body fossils resulted from
loss of water from the soft-bodied invertebrates (some of
them, for instance medusoids, might have contained about
97% water, like recent coelenterates). Results of decompo-
sition of the soft tissues have been interpreted as primary
features of the organisms, leading to speculations about high
surface/volume ratio, osmotic modes of feeding, and low
oxygen content of the Vendian atmosphere.

BATHYMETRIC ZONATION OF VENDIAN
BODY AND TRACE FOSSILS

More than twenty Vendian sequences containing the Edia-
cara-type faunas are known from all over the globe. Every
sequence contains fossiliferous members in a special kind
of facies which occupies a small part of the section. As a
rule, the taxonomic composition of the fossils does not
change much throughout the sequence. A few distinct fossil
assemblages can be recognized which have peculiar elements
that are rare or unknown in other Vendian metazoan locali-
ties, though the distinct character of these fossil biotas could
be explained by taphonomic, paleoecological, and/or paeleo-
biogeographic differences between the compared assem-
blages. This may be the reason why the oldest metazoans
are believed not to manifest any pronounced evolutionary
change throughout the Vendian Period.

The only exception discovered so far is the Vendian se-
quence of the White Sea Region, north of the Russian Plat-
form, where a few distinct faunal assemblages can be ob-
served (Fedonkin, 1981, 1987). Paleogeographically the
White Sea region was open to the ocean during most of the
Vendian Period (Keller and Rozanoyv, 1980; Sokolov and
Fedonkin, 1990), so that the normal marine shallow-water
environments favorable both for animal life and for the pres-
ervation of soft-bodied metazoans existed during a period of
time long enough to embrace the major events in the history
of the biota.

Numerous boreholes drilled in the White Sea region, vol-
canoclastic members as markers, and abundant acritarchs
have made it possible to correlate the separate outcrops (the
major source of the metazoan fossils) with the whole section
of the Vendian. The correlation has led to the succession of
the assemblages, which have some species in common, but
also contain some forms known from Vendian localities else-
where in the world. It turned out that elements of the Nama
(Namibia) fossil assemblages appear earlier in the section
(e.g., Preridinium), and that elements of the Avalon (New-
foundland) biota appear later (e.g., Charnia). Finally, neither
of these occur together — nor do they appear at the higher
stratigraphic levels where the typical elements of the Edia-
cara fauna (South Australia) still exist (e.g., Dickinsonia and
Tribrachidium, which seem to have the longest stratigraphic
ranges in the White Sea region). This succession was re-
vealed early in the course of discovery of the major fossil
localities on the White Sea coast (Fedonkin, 1977, 1981).
However, it was clear that some forms existed through a
period of time embracing the range of the three faunal as-
semblages and, what may be more important, the three as-
semblages are preserved in very different sedimentary facies.

his circumstance led us to consider whether the distribu-
tion of the Vendian fauna was under taphonomice or paleoe-
cological control or some combination of both. These are
virtually undeveloped aspects of Precambrian paleozoology.
If we take into account the body fossils of the benthic or-
ganisms and the trace fossils which are preserved in situ,
then we are able to see the following bathymetric zonations

MIKHAIL A. FEDONKIN

of the Vendian fossils from the shallow water environments
down to the deeper ones:

a) body fossils: Nemiana - Ediacaria - Charnia

b) trace fossils: Skolithos - Palaeopascichnus - Nenoxites

Though zonations of the body fossils and the trace fossils
contain three index names, each name represents an assem-
blage. An important peculiarity of the bathymetric distribu-
tion of the Vendian metazoans was decreasing diversity and
decreasing body size of the vagile benthos offshore.

Global ecosystem restructuring in the Neoproterozoic

Taking into consideration the fundamentally aerobic nature
of all eucaryotes (Margulis et al., 1976) and the growing
body of evidence that at least in animals aerobic metabolism
arose only once and has been strongly conserved throughout
the history of life (Mangum, 1991), the origin and evolution
of the early metazoans could not have been unaffected by
the rise of free oxygen levels in the atmosphere and hydro-
sphere. Oxygen, in turn, being involved in the carbon cycle
of the planet may have an indirect relationship to the climatic
evolution of the biosphere.

The rise of metazoans during the Late Proterozoic and the
incorporation of animals into pre-existing global ecosystems
may have been accomplished under extremely severe con-
ditions. Most of the Precambrian fossil record presents evi-
dence of a stable and conservative ecosystem, driven pre-
dominantly by procaryotic organisms. Both the procaryotic
biota and the environment may have put serious barriers in
the evolutionary pathways of the first multicellular animals,
and even the unicellular eucaryotes as well. Recent pro-
caryote-dominated ecosystems show no evidence of free eco-
logical space for most eucaryotic organisms. The appearance
of metazoans was impossible until the globally-dominating
environments changed and the role of the procaryotes as the
dominants decreased in favor of the eucaryotes. Radical re-
structuring of the global ecosystem from the archaic to the
hew one was accompanied by rise of the eucaryotic trophic
pyramid above the network of biogeochemical interactions
between microbial communities. Ina sense, all the cucaryotic
organisms are an epiphenomenon on the procaryotic back-
ground.

Over a period of almost three billion years the bacterial
communities removed great volumes of various chemical ele-
ments from the hydrosphere and buried them in the sedi-
ments. On the other hand they injected great volumes of the
byproducts of their life activities into the environment. These
long-term processes have resulted in irreversible climatic and
environmental changes. For instance, the oxygen content of
the atmosphere rose due to photosynthesis, at first by cy-
anobacteria and later by eucaryotic algae. Another example
is the withdrawal of carbon dioxide from the atmosphere by
the cyanobacteria and the subsequent conservation of carbon
buried in biogenic deposits (carbonaceous shales, stroma-
tolites, etc.). As a consequence the greenhouse effect was
reduced to produce a very sensitive balance about 850 million

OLDEST FOSSIL ANIMALS

years ago. Since that time glacial periods became more or
less regular events in Earth history. In such a cold episode
we should seek the first appearance of metazoans.

A procaryotic barrier on the metazoan evolutionary
pathway

An important biotic factor which might have allowed ex-
plosive development of animal life and global expansion of
the metazoans in the Vendian basins was the decline of the
stromatolite communities during the Neoproterozoic. This
decline was especially rapid after 850 million years ago.
Vologdin (1962) suggested that the cyanobacteria responsible
for stromatolite formation in the Precambrian may, like their
Recent counterparts, produced toxins. According to his hy-
pothesis, these toxins inhibited animal evolution in its early
stages.

Studies on extant cyanobacteria have revealed a wide va-
riety of toxins (neurotoxins, hepatotoxins, cytotoxins, and
others) which may be directly lethal to small metazoans and
zooplankton, or may reduce their size and number if the
offspring feed on the cyanobacteria (Carmichael, 1994).
These toxins, as well as the marginal marine environments
with high salinity and fluctuating moisture, seem to be the
major factors controlling distribution of metazoan grazers
utilizing the cyanobacteria. The Recent bacterial communi-
ties of sabkhas, marshes and lagoons of arid climatic zones
might be considered analogues for the Precambrian stroma-
tolite communities, which were globally dominant through-
out the major portion of Precambrian life history (Awramik,
1984; Knoll, 1985). The faunas that graze in microbial mats
are of very low diversity (Farmer, 1992). By comparison to
flat bacterial mats, the hemispherical to columnar Proterozoic
stromatolites had greater surface area available for contact
and interaction between the bacterial communities and their
surroundings. This circumstance could have resulted in an
even higher concentration of the toxins in the Proterozoic
shallow water environments.

The idea that grazing and bioturbation by the early meta-
zoans might have caused the decline in stromatolite diversity
and abundance during the Late Proterozoic (Garret, 1970a,
1970b; Awramik, 1971, 1981; Walter and Heys, 1985) seems
to be inconsistent with observations on Recent bacterial mats.
Dominant grazers on modern microbial mats are insects and
crustaceans, but arthropods are unknown from before 550
million years ago (Farmer, 1992). Because most of these
grazers are of small body size (less than a few millimeters),
they do not prevent the development of the bacterial mats,
but rather co-exist with the bacterial communities.

Thus we have to place the first animals into habitats dif-
ferent from those controlled by procaryotic organisms, in
particular, by cyanobacterial communities. Primary biotopes
of the metazoans should be cold enough to inhibit the stro-
matolite-forming cyanobacterial communities. Prior to the
African Glacial Era, which began about 850 million years
ago, cold aquatic environments could exist in a relatively
limited area of the polar regions. Begining with the first in
the series of four glacial periods during the last 300 million

years of the Precambrian, cold water biotopes became more
widespread.

The catastrophic decline of the stromatolite communities
after having dominated the global benthos for almost two
billion years may have been caused by a heterogeneous com-
bination of factors, such as the geochemical, climatic and
paleogeographic consequences of long term activity of pro-
caryotic ecosystems and the rise of eucaryotic algae com-
peting with cyanobacteria for nutrients, habitats and light.

The last revenge of the procaryotic communities on the
eucaryotic ones may have taken place at the very end of the
Vendian. This episode took place after the period of global
expansion of the Ediacara fauna. In the northern part of the
Siberian Platforms it is marked by thick stromatolite carbon-
ates above the beds with a soft-bodied fauna (Sokolov and
Fedonkin, 1984). It is interesting that the black cherts from
the uppermost Vendian, which preserved abundant bacterial
microfossils, do not contain any objects that could be inter-
preted as fossil metazoans. This fact may be interpreted in
favor of the hypothesis of separate biotopes for the oldest
metazoans and the widespread cyanobacterial communities.
That could be true for the first eucaryotes as well.

The cold cradle of animal life

Contrary to those hypotheses that postulate a warm aquatic
environment for the rise of the oldest organisms, | propose
that the first animals, at least, arose in a cold aquatic envi-
ronment. Cold waters of the Recent ocean, in particular the
Antarctic basins, are characterized by some peculiarities that
they may share with the Vendian basins inhabited by the
Ediacara-type fauna. These characteristics are as follows:

1) silicoclastic sedimentation and no carbonates

2) pronounced vertical circulation of the water

3) high concentration of phosphates, nitrates, and other me-
tabolites

4) better aeration of the water

5) less transparent water

6) contrast between meager life on the continent and abun-
dant life in the water

7) high total biomass of the living organisms, especially
plankton

8) higher bioproduction by primary producers

9) short trophic chains

10) low biotic diversity

11) low stability of the biocoenoses

12) dominance of herbivorous planktotrophs among
planktonic metazoans

13) low proportion of predators in the plankton — which
may, however, have undergone population explosions during
some seasons

14) large proportion of coelenterates (medusae and
ctenophores) among the predators

15) dominance of attached forms among the benthic
invertebrates

16) dominance of soft-bodied forms in the benthic com-
munities

17) larger body size than in the forms of the same species
living in warmer waters

18) dominance of forms with direct development (no pelagic
larval stage)

19) seasonal variability in light regime and resource
limitation

20) low rates of metabolism and growth

21) long duration of individual life

22) long generation time

23) low rate of biological degradation.

(Zernov, 1934: Borogov, 1974; Margalef, 1977; Lipps
and Hickman, 1982; White, 1984; Berezina, 1984;
Pearse, McClintock and Bosch, 1991; Stanley and Her-
wig, 1993.) It is remarkable how many of these char-
acters can be seen in the Precambrian metazoans and
their environments.

| should reiterate that most of the fossil localities of the
Vendian fauna occur in the silicoclastic facies accumulated
in moderately cold climatic zones. Thus the Vendian fauna
may well have had all the advantages of the cold water habi-
tats plus one more important factor which used to be over-
looked: an absence of the stable bacterial ecosystems forming
stromatolities.

How representative is the Precambrian fossil record?

The absence of a hard, mineralized skeleton in most of
the Vendian metazoans should make them more or less equal
with respect to preservation potential. Under conditions fa-
vorable for the preservation of soft-bodied animals the num-
ber of species collected should reflect the real biotic diversity,
and the number of fossils should correspond proportionally
to the abundance of each particular species. These expecta-
tions, however, are not confirmed by what we see in the
field nor in the extensive collections of the Vendian fauna.

Though many species of the Ediacara fauna are globally
distributed, their proportion in each particular fossil locality
sometimes differs. Some facies of the Vendian sediments
are characterized by mass preservation of one or two species.
Good examples are the beds with thousands of polyp-like
Nemiana, which are widespread in the Vendian of the
Ukraine, White Sea, Northern Siberia and the Northwest Ter-
ritories of Canada. Selective preservation in this case has to
do with the mode of life of these sedentary forms.

Although the number of specimens of Vendian animals
that have been collected throughout the globe is close to
10,000, their taxonomic diversity remains very low. Over
215 fossil species from the entire Vendian fauna have been
described, but less than half of these taxa have proven tax-
onomically valid (Fedonkin, 1987; Runnegar and Fedonkin,
1992). Even adding about 25 forms of trace fossils which
have been preserved along with the imprints of the soft-
bodied animals leaves the diversity of Vendian metazoans
at a very low level. This fact and the diversity at higher
taxonomic levels lead us to suspect that the Vendian fossil
record reveals only a very small portion of the actual biotic
diversity of the fauna.

MIKHAIL A. FEDONKIN
An obscure radiation

The Varanger glaciation (0.65—0.62 billion years ago) was
the most intensive one in the series of glacial episodes that
took place in the Late Proterozoic (Chumakov, 1968), as is
confirmed by the distribution of tillites of this age all over
the globe, Glacial periods have been accompanied by strong
environmental shifts, including radical climatic and geo-
graphic changes. For instance, marine environments have
been affected by the vast shelf glaciers and floating ice, sharp
fluctuations in sea level, decreasing area of shallow water
habitats of the shelf down to a small strip at the edge of the
platforms, growing temperature gradients, climatic and geo-
graphic isolation, increasing frequency of storm events, etc.
(Fedonkin, 1987). Glacio-eustatic regressions greatly reduced
the area of benthic shallow water habitats, and that may have
had a negative effect upon the marine biota as a whole. The
importance of the shallow water habitats for the Late Pro-
terozoic ecosystem may be underlined by the fact that 83°
of the total benthic biomass in the Recent ocean is concen-
trated on 8% of the substrate, i.e., on the shelf (Leont’ev,
1982). During the Varanger glaciation life was concentrated
in the pelagic waters, which were nutrient-poor and ecologi-
cally homogeneous. That was a prelude to the Vendian ra-
diation of the metazoans. The Post-Varanger glacio-eustatic
transgression over the continents saw a rapid radiation of
megascopic soft-bodied invertebrates known as the Ediacaran
fauna.

Along with the new born taxa of high taxonomic rank
which may have appeared during the early Vendian radiation
of the metazoans (some of which persisted into the Phanero-
zoic) the Ediacaran fauna may well have also included some
pre-Vendian relicts. Even though the time span and the stra-
tigraphic range of the Vendian fauna may be pretty long, as
of now it does not document any prominent sequence in the
appearance of major fossil groups. This fact leaves a few
possibilities open: |) the post-Varanger radiation of metazo-
ans was very rapid, and what we see 1s a stage of evolutionary
stasis accompanied by global expansion of the new born taxa,
or 2) the invertebrates had a long cryptic period of earlier
(pre-Vendian) history, but remained of small body size and
thus cannot be discovered by traditional techniques. Early
metazoans may have lived in environments that have not yet
been studied.

A scenario for the metazoan radiation can be constructed
by the same methods that are used in comparative anatomy
of Recent invertebrates for the study of the phylogenetic re-
lationships and origins of the major groups. Taking into ac-
count all the limitations of the fossil material and the low
diversity of the known Vendian metazoans, we attempt to
reconstruct a very general sketch of their history prior to
their sudden world-wide expansion. The reconstruction
should be based upon an analysis of the diversity at the level
both of the entire biota as well as within major taxonomic
groups, paying special attention to the body plan, mode of
growth, and those structures that can be indentified in later
invertebrates.

Colonization of the shallow water epiplatform basins by

OLDEST FOSSIL ANIMALS

the metazoans required time for adaptation to new environ-
ments, increase in number of individuals, and formation of
populations. All these processes may have been accompanied
by an adaptive radiation which would seem to have taken
place over a geologically brief period of time just before the
moment when the Ediacara fauna expanded globally. This
world-wide expansion of the metazoans marks a period of
their maximum abundance in the Precambrian, and, simul-
taneously, of evolutionary stasis.

General characteristics of the Vendian fauna

The Vendian taphonomic window allows us to envision
the world of the oldest animals represented by their body
fossils as follows: 1) large body size or even gigantism (com-
pared to the small, shelly fossils of the Cambrian); 2) weak
sclerotization of the integument, absolute dominance of soft-
bodied forms, and extreme rarity of organisms with a mineral
skeleton: 3) high diversity of life forms; 4) an ecologically
diverse community, with sedentary forms, vagile benthos,
nekton, plankton and pleuston: 5) relatively few infaunal
metazoans, active filterers and scavengers; 6) dominance of
seston-feeding sedimentators and detritivores; 7) population
density at its maximum in the upper sublittoral zone; 8) short
trophic chains; 9) a high proportion of cosmopolitan species:
10) no bite marks or regenerated structures in the body fos-
sils; 11) high diversity in body plans; 12) low diversity at
the species level; 13) dominance of diploblastic over the
triploblastic organisms both in number of species and in num-
ber of individuals; 14) no clear sequence in the appearance
of the major fossil groups in most Vendian sections.

Ichnological record of the Vendian metazoans

The possibilities for a taxonomic interpretation of the trace
fossils are limited, for with rare exceptions it is difficult to
identify the organisms that produced them, not just at the
species level, but even to taxa of higher rank. Nonetheless
our knowledge of the oldest metazoans is too limited to 1g-
nore these fossils. The well-known metaphor of “fossil be-
havior” applied to the trace fossils does not exhaust the great
potential of paleoichnology for the study of early metazoan
evolution. Being preserved in situ, most trace fossils contain
some information on the environment of metazoan life ac-
tivity and on the effect of the animals on the sedimentary
fabrics. Careful study of the trace fossils yields valuable in-
formation about the producer of the trace: mode of locomo-
tion, feeding habits, behavioral patterns, body morphology,
taxis sensitivity, some physiological functions, etc.

A general overview of the Vendian ichnological record
reveals the following:

1) normally no correlation between the body fossils and
the trace fossils; 2) grazing trails, crawling trails, feeding
burrows and dwelling burrows (in descending order of abun-
dance and diversity); 3) basically “two-dimensional” behav-
ioral stereotypes (meandering predominates); 4+) maximum
diversity in shallow water facies; 5) decreasing diversity and
size in deeper facies; 6) shallow penetration into the sedi-

39
ment; 7) body size of the producers less than that of Cam-
brian counterparts; 8) locomotion dominated by peristaltic
modes of locomotion (by pedal waves or hydrostatic skele-
tons); 9) widespread feeding on small food particles or crop-
ping on benthic microorganisms; 10) coexistence of different
ethological groups in the shallow water ichnocoenoses; 11)
weak bathymetric zonation; 12) low biological processing of
sediments; 13) at the very end of the Vendian and into the
Early Cambrian, increasing body size and diversity of the
behavioral patterns, deeper penetration into the sediment, and
more intensive biological processing of the sediment.

History of the Vendian metazoans

Metazoan diversification seems to have been very rapid
in the Vendian. A period of relatively slow evolution or even
stasis in the middle part of the Vendian (Redkino time) was
followed (Kotlin time) by mass extinctions of many meta-
zoans and decrease in body size of others.

The Vendian metazoan communities may have been ex-
tremely vulnerable (Fedonkin, 1987). The stability of eco-
systems used to be correlated with high biotic diversity and
long trophic chains, but this seems not to be the case for
the Vendian faunas. The shortest trophic chains in the Recent
ocean are found in regions with high phytoplankton produc-
tivity. Those biocoenoses with short food chains produce a
huge biomass but are of very low stabilty (Berezina, 1984).
An additional factor that might have increased the vulner-
ability of the Vendian fauna could be a low replacement rate
of generations in the groups of animals of large body size
and consequently longer life span. According to the concept
of selective extinction (Lewin, 1982), the species charac-
terized by larger body size and slower generation replace-
ment are affected more intensively by extinction.

The decline in taxonomic diversity of the phytoplankton
and the increase in buried vendotaenian algae in the sedi-
ments of the Kotlin basins (Sokolov and Iwanowski, 1990)
may indicate eutrophication of the shallow marine environ-
ments, which, by analogy with present day eutrophic basins,
may have acted as a selective factor in favor of forms having
small body size.

An hypothesized miniaturization in some metazoan groups
close to the end of the Vendian (Fedonkin, 1987) can be
indirectly confirmed by the fact that the first metazoans with
skeletons to undergo an explosive radiation in the early Cam-
brian were indeed very small creatures (Rozanov and
Zhuravlev, 1992; Bengtson and Conway Morris, 1992).

The second increase in body size of invertebrates took
place at the very end of the Vendian (Rovno Stage of Russian
nomenclature). This event, observed in the paleoichnological
fossil record, coincides with the increasing colonization by
invertebrates of the bottom sediments of shallow water en-
vironments, and was immediately followed by the rise of
small shelly organisms (Fedonkin, 1990). Trace fossils of
the Rovno Stage are more diverse, larger, and deeper than
those which are known from the earlier Redkino and Kotlin
Stages. Active colonization of the sediment by the vagile
benthos, as well as new strategies of bioturbation, facilitated

40,

the aeration and decomposition of buried organic matter (Fe-
donkin, 1987; Droser and Bottyjer, 1988).

Actually the Vendian-Cambrian transition was the time of
a rapid vertical expansion of the habitats of benthic meta-
zoans due to the colonization of the deeper sediment below
the surface of the substrate and also to the appearance of
skeletons and reef-like structures that supported organisms
well above the bottom. Such habitat expansion increased the
diversity of the microenvironments both within the sediment
and on the greatly-enlarged surface of such biogenic struc-
tures as reefs and exoskeletons. Growing environmental di-
versity in the benthic realm may have given additional op-
portunities for an adaptive radiation of the new born Early
Cambrian groups of invertebrates.

Our ability to reconstruct the early evolution of the meta-
zoans on the basis of what we know from the Vendian fossil
record is strongly dependent upon opinions concerning the
systematic position of the fossils known as the Ediacara
fauna. These opinions, however, are widely divergent due
to the variety of approaches to the interpretation of these
Precambrian fossils. Consequently there are incompatible
models for early metazoan evolution.

Here I will only mention the non-metazoan interpretation
of the Vendian body fossils (Seilacher, 1984, 1989, 1994;
Norris, 1989; Bergstrom, 1991), Grouping all of the macro-
scopic fossils into one constructional model (the Vendozoa,
later renamed as Vendobionta) is based on schematic, two-
dimensional reconstructions and an inadequate understanding
of taphonomic processes. Invention of the Vendobionta as a
special kingdom or as syncytial protists does not seem nec-
essary. Critiques of the Vendobionta concept from the posi-
tion of taphonomy, comparative anatomy, and phylogenetics
have been provided in several recent works (Gehling, 1991;
Valentine, 1992; Fedonkin, 1992, 1994; Jensen, 1993; Con-
way Morris, 1993).

According to the classical approach developed mainly by
Australian paleontologists, the Precambrian fauna should be
placed into the following taxa: phylum Coelenterata (classes
Hydrozoa, Anthozoa, Scyphozoa, and Conulata, medusae of
uncertain systematic position, and problematic Petalonamae),
phylum Annelida (class Polychaeta), phylum Arthropoda (su-
perclass Trilobitomorpha or Chelicerata), phylum Pogono-
phora, and phylum Echiurida, with some forms treated as of
uncertain systematic position even at the level of the phylum
(Glaessner, 1984).

More recent classifications of the Vendian fauna (Fe-
donkin, 1987, 1991, 1992; Jenkins, 1992; Runnegar, 1992;
Runnegar and Fedonkin, 1992) include some extinct classes
and phyla in addition to extant phyla. This change in the
systematics reflects growing realization that the Vendian
metazoans represent a very peculiar mixture of organisms.
Some of them can be considered relicts of those groups which
appeared long before the Vendian, while the other (newborn)
{axa may represent the ancestors of groups that evolved suc-
cessfully throughout the Phanerozoic.

Body plan analysis of the major groups of the Vendian
fauna (Fedonkin, 1983, 1985, 1991, 1992) has led to a few
important conclusions concerning the systematics and early

MIKHAIL A. FEDONKIN

evolutionary history of the metazoans. Domination of coe-
lenterate-grade organisms in the Vendian may mean that at
a very early stage in evolution the whole range of diversity
of the metazoans could be realized at the diploblastic level
and this diversity in respect of body plans might be greater
than it is at present. In this sense the two phyla Cnidaria
and Ctenophora look, indeed, like relics of the first Metazoa
(Rieger, 1994).

Morphological evolution of the coelenterates seems to have
followed a path from forms with a concentric body plan to
organisms with a radial organization of the body. Lineages
of radially-arranged coelenterates evolved from forms with
non-regulated growth of numerous antimeres to forms with
regulated growth and stable symmetry (Fedonkin, 1985,
1992). Teratological deviations from characteristic symmetry
in some extant groups may be interpreted as illustrating the
wider range of morphological potential that could be realized
in the geological past.

Early coelenterates seem to have evolved from polypoid
life forms with a sedentary mode of life, passive suspension
feeding, and predominantly asexual reproduction — into
medusoid life forms with a free-swimming mode of life, more
sophisticated feeding behavior, sexual reproduction, and
more complex morphology of the gastrovascular and repro-
ductive systems.

Body plan analysis of the Vendian metazoans which can
be interpreted as Bilateria (Triploblastica) seems to indicate
that the evolutionary formation of bilateral symmetry and of
metamerism are closely related, at least in some metazoan
lineages. Hardly expected was some evidence in favor of
cyclomeric theories of the origin of metamerism (Fedonkin,
1985).

Metazoan impact upon the global environment

An explosive radiation of invertebrates was accompanied
by the appearance of new physiologies that could strongly
affect the environments of the Late Protozoic and the Early
Cambrian. This influence could be especially strong during
the periods of growing metazoan abundance. The most im-
portant consequences for the biosphere as a whole may have
resulted from the following phenomena connected with the
life activities of the metazoans:

1. Bioturbation of the sediment resulted in its better aera-
tion, which, in turn, allowed the progressive colonization of
the sediment beneath its surface by a wider variety of aerobic
organisms. Both aeration and increasing life activity within
the sediment promoted the recycling of metabolites in marine
ecosystems. On the other hand, the bioturbation disrupted
the substrate stability that is necessary for the formation of
such biogenic structures as the stromatolites.

2. Biomineralization has resulted in the formation of bio-
clastic deposits and in the creation (with other non-metazoan
sroups of organisms) of the reef as a mechanically stable
biotope and a special ecosystem with a great diversity of
habitats.

3. Filtration of the ocean water by actively filtering organ-
isms has a great impact on the global ocean ecosystem. The

OLDEST FOSSIL ANIMALS

rise of active suspension feeding or filtering in metazoans
at the beginning of the Cambrian has radically changed the
properties of the sediment and water habitats (Fedonkin,
1987, 1992).

The study of filtering in marine planktonic crustaceans
(Vinberg, 1967) has revealed that during 24 hours | milhi-
gram of living weight of the organism is able to filter 360
milliliters of water. Calculations made by Bogorov (1974)
indicate that a volume of water equal to that of the entire
world ocean gets filtered within just half a year! The most
densely inhabited portion of the ocean water (0-500 meters
in depth) is filtered by the organisms in 20 days. To provide
one concrete example, Dankers (1993) estimates that on av-
erage the western Dutch Wadden Sea contains 294 = 10° kg
of mussels (fresh weight). This population would pump 920
x 10° m? of water every day. The western Wadden Sea con-
tains 4,500 x 10° m? of water at low tide. This is a volume
that would be biologically cleared during about five days by
the mussels alone, but there are other filtering groups as
well. An important aspect of such biofiltering is that undi-
gested fine particles packed into fecal pellets settle to the
bottom far more rapidly than do separate fine particles sus-
pended in the sea water.

The rise of active filter-feeding organisms such as sponges,
brachiopods, and many mollusks, arthropods and echino-
derms in the Early Cambrian should have made the ocean
water clear and the photic zone deeper, thus providing op-
portunities for photosynthesizing organisms to occupy deeper
levels both in the water column and on the bottom. The
expansion of the photic zone thus could have resulted in
better oxygenation of the pelagic and benthic habitats via
the activity of the chlorophyll-bearing organisms.

Removal of fine particles from the sea water and packing
them into pellets should have increased the permeability of
the sediment, leading to a better aeration and colonization
of the subsurface bottom environments and to more rapid
oxidation of the buried organic carbon. These factors simul-
taneously reduced the preservation potential of the soft-
bodied metazoans.

4. Increasing length of trophic chains during the Vendian
and even more so in the Cambrian decreased the loss of
major biophile elements and of energy from the ecosystems
because of more efficient biological recycling. That could
lead to global oligotrophication of oceanic waters. This hy-
pothesis is consistent with the general decrease in buried
organic carbon during Early Cambrian time (Knoll, 1992)
as well as by the radiation of Early Cambrian phytoplankton
having external processes, spines, ornamentation and very
small cell size (Yankauskas, 1989). All these morphological
peculiarities of the Early Cambrian phytoplankton can be
interpreted as means to develop a very large surface-volume
ratio that would give some advantage in an oligotrophic en-
vironment (Fedonkin, 1987, 1993). One should note that 70°
of the biomass and 80% of the chlorophyll in the oligotrophic
waters of the Recent ocean belong to the picoplankton.

Some of the feeding habits that were so common in the
Vendian metazoans, such as the passive uptake of food par-
ticles by many sedentary suspension feeders, were rendered

ineffective by the oligotrophication of the Early Cambrian
ocean. This factor may have caused the elimination of some
Ediacara species from the shallow marine habitats in the
Early Cambrian, or even slightly earlier.

Conclusion

The ecological aspect of early metazoan evolution is of
key importance for an understanding of the states of the bio-
sphere during the Late Proterozoic. Very much remains un-
clear. Due to the uncertain systematic position of the oldest
animals and their, evidently, unfamiliar and perhaps extinct
physiologies, a taxon-free approach to the paleoecological
reconstruction as well as non-biological techniques can be
especially effective. Any phylogenetic models based on the
classical neontological or recent molecular data must be put
into, and thus checked by, the paleoecological context which
is a time related aspect of geological history. The problem
of the metazoan origin in the procaryote-dominated (and, in
some extent, antagonistic) world makes us search for a kind
of a contemporary “parallel realm” controlled by the eu-
caryotes. The idea of the “cold cradle of animal life” is an
attempt to solve this problem. However, we have to face a
rather hard question concerning the metazoan colonization
of the habitats historically belonging to communities domi-
nated by procaryotes for at least 3 billion years. An essential
part of this story should be the role of the eucaryotic uni-
cellular organisms (algae, protozoans and fungi), which is
still unknown in detail in spite of the great recent progress
in the study of the Late Proterozoic microfossils. The impact
of the metazoans upon the global environment and ecosystem
was in fact stronger and far more diverse than one caused
by their direct action, such as bioturbation, filtration, biomin-
eralisation or respiration. After their first appearance, meta-
zoan species reached a very high density of population,
which in some Vendian marine biotops was already compa-
rable to the recent one. Thus, even at the level of a relatively
low species diversity the oldest metazoans had become an
important factor regulating the biomass and consequently the
environmental functions of the primary producers and other
organisms at the lower levels of the trophic pyramid.

Acknowledgments

The author is grateful to Drs. Michael Ghiselin, Giovanni
Pinna and Marvalee Wake for organizing the conference and
providing hospitality. This paper is part of project 9305-05-
8816 supported by the Russia Foundation for Fundamental
Research. My special thanks are due to Dr. Ghiselin and to
Dr. Ellis Yochelson for critical discussion and careful editing
of the manuscript.

Literature Cited
Andres, D. 1989. Phosphatisierte Fossilien aus dem unteren Ordoviz

von Sudschweden. Berliner Geowissenschaftliche Abhandlungen
106:9-19.

Awramik, S. M. 1971. Precambrian columnar stromatolite diversity:
reflection of metazoan appearance. Science 174:825—827.

— —. 1981. The pre-Phanerozoic biosphere: three billion years
of crises and opportunities. Pages 83-102 in M. H. Nitecki, editor.
Biotic Crises in Ecological and Evolutionary Time. Academic
Press, New York.

. 1984. Ancient stromatolites and microbial mats. Pages 1—
22 in Y. Cohen, R. W. Castenholz and H. O. Halvorson, editors.
Microbial Mats: Stromatolites. Alan R. Liss, New York.

Bengston, S. and S. Conway Morris. 1992. Early radiation of
biomineralizing phyla. Pages 447-481 in J. H. Lipps and P. W.
Signor, editors. Origin and Early Evolution of the Metazoa. Ple-
num Press, New York.

Berezina, N. A. 1984. Hydrobiology, 4th Edition. Legkaya 1
Pichevaya Promyshlennost, Moscow. (In Russian.)

Bergstrom, J. 1991. Metazoan evolution around the Precambran-
Cambrian transition. Pages 25—34 in A. M. Simonetta and S. Con-
way Morris, editors. The Early Evolution of Metazoa and the
Significance of Problematic Taxa. Cambridge University Press,
Cambridge.

. 1994. Ideas on early animal evolution. Pages 460-466 in
S. Bengtson, editor. Nobel Symposium, No. 84: Early Life on
Earth. Columbia University Press, New York.

Bogoroyv, V.G. 1974. Plankton of the World Ocean, Nauka, Mos-
cow. (In Russian.)

Bonner, J. T. 1965. Size and Cycle: an Essay on the Structure of
Biology. Princeton University Press, Princeton.

1988. The Evolution of Complexity by Means of Natural
Selection. Princeton University Press, Princeton

Carmichael, W. W. 1994. The toxins of the Cyanobacteria. Scien-
tific American 1994:78—86

Chapman, D. J. 1992. Origin and divergence of protists. Pages 477
483 in JW. Schopt and C. Klein, editors. The Proterozoic Bio-
sphere: a Multidisciplinary Study. Cambridge University Pres,
Cambridge

Chen, Jun-Y. 1988. Precambrian metazoans of the Huai River drain-
age area (Anhui, E. China): their taphonomie and ecological evi-
dence. Lethaia 69:189-215

Christen, R. 1994. Molecular phylogeny and the origin of Metazoa.
Pages 467-474 i S. Bengtson, editor. Nobel Symposium, No
S4: Early Life on Earth. Columbia University Press. New York.

Christen, R., A. Ratto, A. Baroin, R. Perasso, K. G. Grell, and A.
\doutte. 199], Origin of metazoans: a phylogeny deduced from
sequences of the 28S ribosomal RNA. Pages 1-9 in A.M. Si-
monetta and S, Conway Morris, editors, The Early Evolution of
Metazoa and the Significance of Problematic Taxa. Cambridge
University Press, Cambridge.

Chumakov, N.M. 1978. Transactions of the Geological Institute,
USSR Academy of Sciences, Volume 308: Precambrian Tillites

MIKHAIL A. FEDONKIN

and Tilloids (Problems of the Precambrian Glaciations). Nauka,
Moscow.

Cloud, P. E., Jr. 1968. Pre-metazoan evolution and the origins of
the Metazoa. Pages 1—72 in E. T. Drake, editor. Evolution and
Environment. Yale University Press, New Haven.

Conway Morris, S. 1993. Ediacaran-like fossils in Cambrian Bur-
gess-Shale type faunas of North America. Palaeontology 36:593—
635.

Damuth, J. D., D. Jablonski, J. A. Harris, R. Potts, R. K. Stucky,
H.-D. Sues, and D. Weishampel. 1992. Taxon-free charac-
terization of animal communities. Pages 183-203 in A. K. Be-
hrensmeyer, J. D. Damuth, W. A. DiMechele, R. Potts and H.-D.
Sues, editors. Terrestrial Ecosystems through Time. University
of Chicago Press, Chicago.

Dankers, N. 1993. Integrated estuaring management — obtaining a
sustainable yield of bivlave resources while maintaining environ-
mental quality. Pages 479-511 in R. F. Dame, editor. NATO ASI
Series, Vol. G33: . Springer Verlag, Berlin.

Des Marais, D. J. 1990. Microbial mats and the early evolution of
life. Trends in Ecology and Evolution 5:140—143.

Dickerson, R. E. 1971. The structure of cytochrome “c” and the
rate of molecular evolution. Journal of Molecular Evolution 1:26—
45.

Droser, M. L. and D. J. Bottjer. 1988. Trends in depth and extent
of bioturbation in Cambrian carbonate marine environments,
western United States. Geology 16:233—236.

Durham, J. W. 1970. The fossil record and the origin of the Deu-
terostomia. Proceedings of the North American Paleontological
Convention 1969 H:1 104-1132.

Eldredge, N. 1989. Macroevolutionary Dynamics: Species, Niches,
and Adaptive Peaks. McGraw-Hill Publishing Company. New
York.

Erwin, D. H. 1989. Molecular clocks, molecular phylogenies and
the origin of phyla. Lethaia 22:251—257

Farmer, J. D. 1992. Grazing and bioturbation in moderm microbial
mats. Pages 295-297 in J. W. Schopf and C. Klein, editors. The
Proterozoic Biosphere: a Multidisciplinary Study. Cambridge
University Press, Cambridge

Fedonkin, M. A. 1981. Transactions of the Geological Institute, 342:
White Sea Biota of the Vendian (Precambrian Non-Skeletal Fauna
of the Russian Platform North). Moscow, Nauka. (In Russian.)

1985. Precambrian metazoans: the problems of preserva-
tion, systematics and evolution, Philosophical Transactions of the
Royal Society of London B311:27-45.

. 1987. Transactions of the Paleontological Institute, N226:
Non-skeletal Fauna of the Vendian and its Place in the Evolution

of the Metazoans, Nauka, Moscow.

———. 1990. Precambrian metazoans. Pages 17-24 in DoE. G

OLDEST FOSSIL ANIMALS

Briggs and P. R. Crowther, editors. Palaeobiology: a Synthesis.
Blackwell Scientific Publications, Oxford.

. 1990. Systematic description of the Vendian Metazoa.
Pages 71-120 in B.S. Sokolov and A. B. Iwanowski, editors.
The Vendian System. Volume 1. Springer-Verlag, Berlin.

. 1992. Vendian faunas and the early evolution of the Meta-
zoa. Pages 87-129 in J. H. Lipps and P. W. Signor, editors. Origin
and Early Evolution of the Metazoa. Plenum Press, New York.

. 1993. Paleobiology of the Precambrian: on the way to the
synthesis. Pages 7-21 in B.S. Sokolov and A. B. Iwanowski,
editors. Faunas and Ecosystems of the Geological Past. Nauka,
Moscow. (In Russian.)

. 1994. Vendian body fossils and trace fossils. Pages 370—
388 in S. Bengtson, editor. Nobel Symposium, No. 84: Early
Life on Earth. Columbia University Press, New York.

. 1996. Geobiological trends and events in the Precambrian
biosphere. Pages 89-112 in O. T. Walliser, editor. Global Bio-
logical Events in Earth History. Springer-Verlag, Berlin.

Fedonkin, M. A. and B. N. Runnegar. 1992. Proterozoic metazoan
trace fossils. Pages 389-395 in J. W. Schopf and C. Klein, editors.
The Proterozoic Biosphere: a Multidisciplinary Study. Cambridge
University Press, Cambridge.

Fedonkin, M. A., E. L. Yochelson, and R. J. Horodyski. 1994. An-
cient Metazoa. National Geographic Research and Education
10:200—223.

Field, K. G., G. J. Olsen, D. J. Lane, S. Giovannoni, M. T. Ghiselin,
E.C. Raff, N.R. Pace, and R. A. Raff. 1988. Molecular phylo-
geny of the animal kingdom. Science 239:748—753.

Garrett, P. 1970. Deposit feeders limit development of stromatolites.
Bulletin of the American Association of Petroleum Geologists
54:848.

. 1970. Phanerozoic stromatolites: non competitive ecologic
restriction by grazing and burrowing animals. Science 169:171—
173:

Gehling, J. G. 1991. The case for Ediacaran fossil roots to the meta-
zoan tree. Memoirs of the Geological Society of India 20:183—
223)

Ghiselin, M. T. 1988. The origin of molluscs in the light of mo-
lecular evidence. Oxford Surveys in Evolutionary Biology. 5:66—
95.

. 1989. Summary of our current knowledge of metazoan
phylogeny. Pages 261-272 in B. Fernholm, K. Bremer and H.
Jomvall, editors. The hierarchy of life: molecules and morphology
in phylogenetic analysis. Exerpta Medica, Amsterdam.

Glaessner, M. F. 1958. New fossils from the base of the Cambrian
in South Australia. Transactions of the Royal Society of South
Australia 81:185—188.

. 1961. Precambrian animals. Scientific American 204:72—

78.

. 1969. Trace fossils from the Precambrian and Cambrian.
Lethaia 2:369-393.

. 1971. The genus Conomedusites Glaessner and Wade and
the diversification of the Cnidaria, Palaéontologische Zeitschrift
43:7-17.

. 1976. A new genus of polychaete worms from the Late
Precambrian of South Australia. Transactions of the Royal So-
ciety of South Australia 100:169—170.

1979. Biogeography and biostratigraphy: Precambrian.
Pages 79-118 in R.C. Moore, R. A. Robison and C. Teichert,
editors. Treatise on Invertebrate Paleontology. Volume A. Geo-
logical Society of America, Lawrence.

. 1980. Parvancorina: an arthropod from the Late Precam-
brian of South Australia. Annalen der Naturhistorisches Museum
Wien 83:83—90.

. 1984. The Dawn of Animal Life: a Biohistorical Study.
Cambridge University Press, Cambridge.

Glaessner, M.F. and B. Daily. 1959. The geology and the late
Precambrian fauna of the Ediacara fossil reserve. Records of the
South Australian Museum 13:369-401,

Glaessner, M. F. and M. Wade. 1966. The Late Precambrian fossils
from Ediacara, South Australia. Palaeontology 9:599-628.

——. 197]. Precambridium: a primitive arthropod. Lethaia 4:71-
THe

Goldring, R. and C. N. Curnow. 1967. The stratigraphy and facies
of the Late Precambrian of Ediacara, South Australia. Journal of
the Geological Society of Australia 14:195—214.

Grey, K. and]. R. Williams. 1990. Problematic bedding-plane mark-
ings from the Middle Proterozoic Manganese Subgroup, Bange-
mall Basin, Western Australia. Precambrian Research 46:307
327.

Hantzschel, W. 1975. Treatise on Invertebrate Paleontology, Part
W: Miscellanea: Supplement |: Trace Fossils and Miscellanea,
2nd Edition. Geological Society of America, Lawrence.

Hoffman, H. J. 1971. Precambrian fossils, pseudofossils and proble-
matica in Canada. Geological Survey of Canada Bulletin 189:1
146.

Hoffman, H. J.. G. M. Narbonne, and J. D. Aitken. 1990. Ediacaran
remains from intertillite beds in northwestern Canada. Geology
18:1 199-1202.

Hofmann, H. J. 1992. Megascopic dubiofossils. Pages 413-419 in
J. W. Schopf and C. Klein, editors. The Proterozoic Biosphere:
a Multidisciplinary Study. Cambridge University Press, Cam-
bridge.

Horodyski, R. J. 1982. Problematic bedding-plane markings from
the Middle Proterozoic Appekunni Argillite, Belt Supergroup,
north-western Montana. Journal of Paleontology 56:882—889.

Jenkins, R. J. F. 1992. Functional and ecological aspects of Edia-
caran assemblages. Pages 131-176 in J. H. Lipps and P. W. Sig-
nor, editors. Origin and Early Evolution of the Metazoa. Plenum
Press, New York.

Jensen, S. 1993. Trace fossils, body fossils, and problematica from
the Lower Cambrian Mickwitzia Sandstone, south-central Swe-
den. Ph.D. Thesis. Institute of Earth Sciences, Historical Geology
and Paleontology, Uppsala University.

Keller, B.M. and A. Y. Rozanoy, editors, 1980. Paleogeography
and Lithology of Vendian and Cambrian of the Western Part of
the East-European Platform. Nauka, Moscow.

Kidwell, S. M., and T. Baumiller. 1990. Experimental disintegration
of regular echinoids: role of temperature, oxygen, and decay
thresholds. Paleobiology 16:247—271.

Knoll, A. H. 1985. A paleobiological perspective of sabkhas. Pages
407-427 in G. M. Friedman and W. E. Krumbein, editors. Hy-
persaline Ecosystems. Springer Verlag, Berlin.

. 1992. Biological and biogeochemical preludes to the Edia-
caran radiation. Pages 53-84 in J. H. Lipps and P. W. Signor,
editors. Origin and Early Evolution of the Metazoa. Plenum Press,
New York.

Lake, J. A. 1990. Origin of the Metazoa. Proceedings of the National
Academy of Sciences USA 87:763—766.

Leont’ev, R. 1982. Physical Geography of the World Ocan. Moscow
State University Press, Moscow.

Lewin, R. 1982. Extinction leaves its mark on ecology. Science
218:42-43

Lipps, J. H. and C.S. Hickman. 1981. Origin, age, and evolution
of Antarcuc deep-sea faunas. Pages 325-359 in W.G. Ermst and
J.G. Morin, editors. The Environments of the Deep Sea. Rubey
Volume 2. Prentice Hall, Englewood Cliffs, Paper from a collo-
quium, Los Angeles, April-June 1979.

Mangum, C. P. 199]. Precambrian oxygen levels, the sulfide bio-
system, and the origin of the Metazoa, Journal of Experimental
Zoology 260:33-42

Margalet, R. 1977. Ecosystem diversity difference: poles and trop-
ics. Pages 367-376 in M. J. Dunbar, editor. Polar Oceans. Arctic
Institute of North America, Calgary

Margulis, L.. J.C. G. Walker, and M. Rambler. 1976. Reassessment
of roles of oxygen and ultraviolet light in Precambrian evolution.
Nature 264:620—-624

McLaughlin, P. J. and M. O. Dayhoff. 1973, Eukaryote evolution:
a View based on Cytochrome c. Journal of Molecular Evolution
2:99-116

Monty, C.L. V. 1974. Precambrian background and Phanerozoic
history of stromatoliic communities, an overview. Annales de la
Societe Geologique de Belgique 96:585- 624.

Miller, KJ. 1983. Crustacea with preserved soft parts from the
Upper Cambrian of Sweden. Lethaia 6:93—109

MIKHAIL A. FEDONKIN

Norris, R. 1989. Cnidarian taphonomy and affinities of the Ediacara
biota. Lethaia 22:381-399,

Patterson, C. 1989. Phylogenetic relations of major groups: conclu-
sions and prospects. Pages 471488 i B. Fernholm, K. Bremer
and H. Jornvall, editors. The hierarchy of life: molecules
and morphology in phylogenetic analysis. Exerpta Medica,
Amsterdam,

Pearse, J. S., J. B. McClintock, and I. Bosch. 1991. Reproduction
of Antarctic benthic invertebrates: tempos, modes, and timing.
American Zoologist 31:65—88.

Raff, R. A., K.G. Field, G. J. Olsen, S. J. Giovannoni, D. J. Lane,
M. T. Ghiselin, N. R. Pace, and E. C. Raff. 1989. Metazoan phy-
logeny based on analysis of 18S ribosomal RNA. Pages 247-260
in B. Fernholm, K. Bremer and H. Jornvall, editors. The Hier-
archy of Life: Molecules and Morphology in Phylogenetic Analy-
sis. Exerpta Medica, Amsterdam.

Rieger, R. M. 1994. Evolution of the “lower” Metazoa. Pages 475—
488 in S. Bengtson, editor. Nobel Symposium, No. 84: Early
Life on Earth. Columbia University Press, New York.

Rozanoy, A. Y., and A. Y. Zhurlavlev. 1992. The Lower Cambrian
fossil record of the Soviet Union. Pages 205-282 in J. H. Lipps
and P. W. Signor, editors. Origin and Early Evolution of the Meta-
zoa. Plenum Press, New York.

Runnegar, B. 1982. Molecular-clock data for the origin of the animal
phyla. Lethaia 15:199-205.

———. 1982, Oxygen requirements, a biology and phylogenetic
significance of the late Precambrian worm Dickinsonia and the
evolution of the burrowing habit. Alcheringa 6:223—239.

--—- 1986. Molecular palaeontology. Palaeontology 20:1—24.
———. 1992. Origin and diversification of the Metazoa. Pages
485 in J. W. Schopt and C. Klein, editors. The Proterozoic Bio-
sphere: a Multidisciplinary Study. Cambridge University Press,
Cambridge

Runnegar, B. N. and M. A. Fedonkin. 1992. Proterozoic metazoan
body fossils. Pages 369-388 iJ. W. Schopf and C. Klein, editors.
The Proterozoic Biosphere: a Multidisciplinary Study. Cambridge
University Press, Cambridge

Schonborn, W. 1987. On the primary biotops of the animal world.
Zhurnal Obshchei Biologi 48:27—40.

Schopf. J. W. and C. Klein, editors. 1992. The Proterozoic Bio-
sphere: a Multidisciplinary Approach. Cambridge University
Press, Cambridge.

Seilacher, A. 1984. Late Precambrian Metazoa: preservational or
real extinctions? Pages 159-168 7 H. D. Holland and A. F. Tren-
dall, editors. Patterns of Change in Earth Evolution. Springer-
Verlag, Berlin.

1989, Vendozoa: organismic construction in the Protero-
zoic biosphere. Lethaia 22:229-239,

OLDEST FOSSIL ANIMALS

. 1994. Early multicellular life: late Proterozoic fossils and
the Cambrian explosion. Pages 389-400 in S. Bengtson, editor.
Nobel Symposium, No. 84: Early Life on Earth. Columbia Uni-
versity Press, New York.

Semikhatov, M. A. and M. E. Raaben. 1993. Dynamics of the sys-
tematic diversity of the Riphean and Vendian stromatolites in
Northern Eurasia. Stratigraphy. Geological Correlation 1:3—12.

. 1994. Dynamics of the global diversity of the Proterozoic
stromatolites. Paper 1: Northern Eurasia, China and India. Stra-
tigraphy. Geological Correlation 2:10—-32.

Smith, A. B. 1989. RNA sequence data in phylogenetic reconstruc-
tion: testing the limits of resolution. Cladistics $:321—344.

Sokolov, B.S., and M. A. Fedonkin. 1984. The Vendian as the
terminal system of the Precambrian. Episodes 7:12—19.

, editors. 1990. The Vendian System. Volume 2. Springer-
Verlag, Berlin.

Sokolov, B.S. and A. B. Iwanowski, editors. 1990. The Vendian
System. Volume 1. Springer Verlag, Berlin.

Sprigg, R. C. 1947. Early Cambrian (?) jellyfishes from the Flinders
Ranges, South Australia. Transactions of the Royal Society of
South Australia 71:212—234.

. 1949. Early Cambrian “jellyfishes” of Ediacara, South Aus-
tralia, and Mount John, Kimberley District, Western Australia.
Transactions of the Royal Society of South Australia 73:72—99.

Stanley, J., and R. P. Herwig. 1994. Degradation of particulate or-
ganic material in the Antarctic. Pages 241—264 in E. I. Friedman,
editor. Antarctic Microbiology. Wiley-Liss, New York.

Sun, W. 1986. Are there pre-ediacaran metazoans? Precambrian
Research 31:409-410.

Sun, W., G. Wang, and B. Zhou. 1986. Macroscopic worm-like
body fossils from the Upper Precambrian (900-700 Ma), Huainan
district, Anhui, China and their stratigraphic and evolutionary
significance. Precambrian Research 31:377-403.

Torsvik, T. H. 1994. Baltic-Siberia connection challenges traditional
tectonic notions. EOS 40:461-462.

Torsvik, T. H., M. A. Smethurst, R. Van der Voo, A. Trench, N.
Abrahamsen, and E. Halvorsen. 1992. Baltica. A synopsis of Ven-
dian-Permian palaeomagnetic edata and their palaeotectonic 1m-
plications. Earth Science Reviews 33:133—152.

Valentine, J. W. 1991. Major factors in the rapidity and extent of
the metazoan radiation during the Proterozoic-Phanerozoic tran-
sition. Pages 11-13 in A.M. Simonetta and S. Conway Morris,
editors. The Early Evolution of Metazoa and the Significance of
Problematic Taxa. Cambridge University Press, Cambridge.

. 1992. Dickinsonia as a polypoid organism. Paleobiology
18:378-382.

———. 1994. The Cambrian explosion. Pages 401-411 in S. Bengt-
son, editor. Nobel Symposium, No. 84: Early Life on Earth. Co-
lumbia University Press, New York,

Vinberg, G.G. 1967. Peculiarities of aquatic ecological systems.
Zhurnal Obshchei Biologi 28. (In Russian.)

Vodanjuk, S. A. 1989, Remains of the non-skeletal metazoans from
the Khatyspyt Formation of the Oleniok Uplift. Pages 61—74 in
V.V. Khomentovski and Y. K. Sovetov, editors. Transactions of
the Institute of Geology and Geophysics, Siberian Branch of the
USSR Academy of Sciences: Late Precambrian and Early Paleo-
zoic of Siberia. Actual Questions of the Stratigraphy. Siberian
Branch of the USSR Academy of Sciences, Novosibirsk. (In Rus-
sian.)

Volgodin, A. G. 1962. Oldest Algae of the USSR. Publishing House
of the USSR Academy of Sciences, Moscow. (In Russian.)

Wade, M. 1968. Preservation of soft-bodied animals in Precambrian
sandstones at Ediacara, South Australia. Lethaia 1:238—267.

. 1971. Bilateral Precambrian chondrophores from the Edia-
cara fauna, South Australia. Proceedings of the Royal Society of
Victoria 84:183—188.

. 1972. Dickinsonia: polychaete worm from the Late Pre-
cambrian Ediacara fauna, South Australia. Memoirs of the
Queensland Museum 16:171—190.

. 1972. Hydrozoa and Scyphozoa and other medusoids from
the Precambrian Ediacara fauna, South Australia. Palaeontology
15:197—225.

Walter, M. R. and G. R. Heys. 1985. Links between the rise of the
Metazoa and the decline of the stromatolites. Precambrian Re-
search 20:149—174

Wang, G.-X. 1982. Late Precambrian Annelida and Pogonophora
from the Huanian of Anhui Province. Bulletin of the Tianjin In-
stitute of Geology and Mineral Resources 6:9-22. ((In Chinese. ))

White, M. G. 1984. Marine benthos. Pages 421-461 in R. M. Laws,
editor. Antarctic Ecology. Volume 2, Academic Press, London.

Yankauskas, T. V., editor. 1989. Microfossils from the Precambrian
of the USSR. Nauka, Leningrad. (In Russian.)

Zernov, S. A. 1934. General Hydrobiology. Biomedgiz, Moscow
(In Russian.)

Zheng, W. W. 1980. A new occurrence of fossil group of Chuaria
from the Sinian System in north Anhui and its geological mean-
ing. Bulletin of the Tianjin Institute of Geology and Mineral Re-
sources 1:49-69.

BIOGEOGRAPHIC CAUSES OF DISCONTINUITY
IN THE FOSSIL RECORD OF THE AMMONITES

Giovanni Pinna

Museo Civico di Storia Naturale
Corso Venezia, 55
20121 Milan, Italy

Introduction

The evolutionary models that palaeontologists create on
the basis of palaeontological material are often dependent
upon theoretical assumptions about the fossil record. These
models differ profoundly, depending upon whether the fossil
record is thought to be complete enough to supply a picture
of the biological past that 1s very close to how events really
took place, or if, on the contrary, the record is considered
to be so full of gaps as to invalidate any palaeontological
evidence with respect to the mechanisms that operated during
evolution.

On the basis of the palaeontolgical data, most phyletic line-
ages seem to have evolved in a discontinuous way. If we
assume that the fossil record is substantially complete, then
the evolutionary process will be subject to interpretation as
just such a discontinuous process. If, on the contrary, we
admit the possibility that the fossil record is incomplete, and
therefore unable to supply a real image of the past, we will
be able to suppose gradual evolution.

The debate on the completeness of the fossil record has
marked the whole history of palaeontological studies on evo-
lution, starting with Lamarck and Cuvier, and has resulted
in two different visions of the evolutionary process. These
can be exemplified by Simpson’s gradualist vision, in which
microevolution and macroevolution are but two aspects of
the same continuous process, and by Schindewolf’s salta-
tionist vision, in which microevolution and macroevolution
represent two intrinsically different processes.

The Reasons for Incompleteness of the Fossil Record

It is well known that fossilization is a selective process
and that its selective power (the so-called taphonomic filter)
depends on several factors, among which the most important
are the nature of the organic remains and the conditions of
the sedimentary environment. It is also known that geological
factors, such as the amount of stratal condensation, affect
the continuity and the completeness of the sedimentary series,
and thereby the completeness of the fossil record.

Nonetheless, other factors exist that are unrelated to the
selective effect of the fossilization process and to the action
of geological events and that, even under the most favorable
environmental and organic conditions, can affect the com-
pleteness of the fossil record, and thus contribute to the ap-
pearance of discontinuities in phyletic lineages. These are

New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

biogeographic factors related to the disjunction of phyletic
lineages due to population movements among adjacent bio-
geographic provinces, and in turn related to the environ-
mental conditions of the provinces.

The aim of this paper is to explain how such biogeographic
factors can produce apparent discontinuity in the fossil re-
cord, something that is difficult to intepret in the absence of
a complete picture of the distribution of the group under
consideration in space and time. As an example, I will discuss
the evolution of the Toarcian (Lower Jurassic) subfamily
Phymatoceratinae, which took place in two adjacent bio-
geographic provinces, and I will show that the discontinuities
in the development of the phyletic lines found in one of the
two provinces are essentially due to biogeographic factors.

Biogeographic Provinces During the Toarcian

So far as the ammonite faunas are concerned, different
faunistic provinces can be identified during the Toarcian.
Among these are a Sub-Mediterranean Province, correspond-
ing to the extension of the epicontinental seas over part of
continental Europe (especially France, Germany and Eng-
land) and a Mediterranean Province, corresponding to the
western part of Tethys basin, sedimentary deposits of which
outcrop especially in Italy, North Africa, Greece, the Bal-
kans, the Caucasus (Fig. 1).

The ammonite faunas of the two provinces have a basically
different structure. On the average, the ammonite fauna in
the Sub-Mediterranean Province has a relatively low taxo-
nomic diversity and a high number of individuals, while in
the Mediterranean Province it has a relatively high taxonomic
diversity and a low number of individuals. Such faunal struc-
ture would indicate, at least as far as the ammonites are
concerned, a r-selective environment for the Sub-Mediterra-
nean Province and a K-selective environment for the Medi-
terranean Province. Hallam (1975) ascribed these faunistic
differences to environmental instability and stability, respec-
tively, in agreement with the model of Sanders (1968, 1969).
The study of Phymatoceratinae has shown that some faunal
mixing among the two provinces took place at different
times. Individuals belonging to mediterranean endemic
groups occasionally dispersed to the Sub-Mediterranean
Province, but without giving rise to continuous phyletic lines;
however, only on two occasions did mediterranean groups
establish permanent populations in the Sub-Mediterranean
Province that gave rise to new phyletic lineages.

eoeeeeeeceece
eoeeeeseee

GIOVANNI PINNA

eeeceeeeeeeeoeeeee:
eeeeceeeeeeeeeee eee

Mediterranean

FIGURE 1. Map of Europe and adjacent regions as they appeared in the Toarcian, showing Boreal (upper left), Sub-Mediterranean (center)

and Mediterranean (lower right) faunistic provinces for ammonites. Land 1s stippled. (After Smith, A. G.,

1994)

Because of the population movements of ammonites be-
tween the two provinces, the different structure of the dif-
ferent environments takes on remarkable importance. In fact,
it is highly probable that most of the population movements
take place from saturated and highly-competitive K-selective
environments to non-saturated, moderately-competitive r-se-
lective environments. In the specific case of the ammonites
and the two provinces considered here, there was a general
tendency for mediterranean groups to invade the Sub-Medi-
terranean Province in successive stages during the Toarcian.

Stratigraphy of Toarcian

Because of the differences in ammonite faunas in the Sub-
Mediterranean and the Mediterranean Provinces, the bio-stra-
tigraphic zonations in the two provinces do not match, being
based on endemic zonal index species in each province.
However, correlation between the two biostratigraphic zona-
tions is possible because of repeated entries of mediterranean

Smith, D. G., and Funnell, B. M.,

transient isolated forms into the Sub-Mediterranean Province
and because migrations into the same Province that twice
established permanent populations. Reference is here made
to the following correlational stratigraphic scheme:

Mediterranean
Province

Sub-Mediterranean
Province
levesquei Zone meneghini Zone

bifrons Zone bifrons Zone

falciferum Zone serpentinus Zone

FOSSIL RECORD DISCONTINUITY
Evolution of the Subfamily Phymatoceratinae

In 1957 Arkell (in Arkell, Kummel and Wright, 1957) con-
sidered the subfamily Phymatoceratinae to be a polyphyletic
group, derived from several members of the Hildoceratidae,
and he ascribed to it eight genera, including Haugia and
Phymatoceras. The genera Denckmannia, Lillia and Char-
tronia, previously established by different authors as separate
taxa, were considered as synonyms of Phymatoceras.

The subfamily Phymatoceratinae is not applied here in the
wider sense of Arkell, but instead more precisely as a mo-
nophyletic group consisting of five phyletic lineages derived
from one another during Toarcian (Pinna and Levi-Setti,
1973). The five phyletic lineages lead to the following groups
of forms, which are considered, partly arbitrarily, as repre-
senting genera:

1) group of Ammonites erbaensis Hauer, 1856, correspond-
ing to the genus Phymatoceras Hyatt, 1867,

2) group of Lillia narbonensis Buckman, 1898, correspond-
ing to the genus Chartronia Buckman, 1898,

3) group of Hildoceras (Lillia) chelussii Parish and Viale,
1906, corresponding to the genus Li/lia Bayle, 1878,

4) group of Denckmannia tumefacta Buckman, 1898, cor-
responding to the genus Denckmannia Buckman, 1898,

5) group of Ammonites variabilis (d’Orbigny, 1844), cor-
responding to the genus Haugia Buckman, 1888.

Gabilly (1976) revised the Phymatoceratinae of the stra-
totype of Toarcian (Thouars Region, France), in the Sub-
Mediterranean Province. He ascribed to the subfamily the
genus Phymatoceras and also the genera Denckmannia and
Haugia, which he supposed were derived from it.

The phyletic relationships derived by Gabilly do not cor-
respond to those of the Phymatoceratinae of the Mediterra-
nean Province, for they were carried out on an intermittent
fossil record. The fossil record of Phymatoceratinae in the
Sub-Mediterranean Province (with the exception of the gen-
era Haugia and Denckmannia) does not reflect the true phy-
logeny of this group of ammonites. Rather, it corresponds
to occasional entries of transient individuals and to migration
populations of separate phyletic lineages that evolved con-
tinuously in the Mediterranean Province.

Such discontinuity in the fossil record had the following
effects:

| — The lack of transitional forms resulted in an erroneous
systematic interpretation. United in the same genus were spe-
cies that, in the light of the different Mediterranean faunas,
would seem to belong to different entities. For instance, in
the systematic scheme drawn by Gabilly the genus Phyma-
toceras includes both Phymatoceras s.str. and representatives
of the genus Chartronia, while the genus Denckmannia in-
cludes both Denckmannia s.str. and representatives of the
genus Lillia.

2—The incompleteness of the fossil record precluded
documentating the cenogenetic and palingenetic mechanisms
by which the evolutionary novelities appeared (observable
only in relatively continuous series) and in turn it prevented
establishing ancestor-descendant links of the different forms.

3— The incompleteness of the documentation, without
any kind of data on the evolution of the group that took
place in the Mediterranean area, prevented the systematic
identification of many forms, resulting in the establishment
of five new systematic entities out of nineteen analyzed spe-
cles.

Comparative study of the Mediterranean and Sub-Medi-
terranean faunas (Pinna and Levi-Setti 1971) made it clear
that the phyletic lineages corresponding to the genera Phy-
matoceras, Chartronia, and Lillia are endemic to the Medi-
terranean Province, while those corresponding to the genera
Haugia and Denckmannia are endemic to the Sub-Mediter-
ranean Province.

Therefore the five phyletic lineages evolved in parallel
from the stem group consisting of the genus Phymatoceras,
which thus became paraphyletic, three in the mediterranean
area and two in the sub-mediterranean area (Fig. 2).

In the Mediterranean Province:

— the genus Phymatoceras originated from the genus
Hildaites at the beginning of brifrons Zone and evolved until
becoming transformed into Catulloceras ssp. at the beginning
of meneghinu Zone,

—the genus Lillia originated from the genus Phymato-
ceras in the middle of bifrons Zone and then died out in the
upper part of erbaense zone,

—the genus Chartronia evolved in the middle of bifrons
Zone from the genus Phymatoceras and then died out in the
upper part of erbaense Zone, or little earlier.

In the Sub-Mediterranean Province:

—the genus Chartronia populated the Sub-Mediterranean
Province in the middle of the bifrons Zone with primitive
forms, identical to coeval mediterranean forms; those forms
established permanent populations that evolved and gave rise
to the sub-mediterranean endemic phyletic lineage of the ge-
nus Haugia,

—the genus Phymatoceras populated the Sub-Mediterra-
nean Province at the beginning of varibilis Zone with forms
identical to coeval mediterranean forms, which established
permanent populations that evolved and gave rise to the sub-
mediterranean endemic phyletic lineage of the genus Denck-
mannia.

As already mentioned, there was a constant trend of the
populations of Mediterranean ammonites to invade the Sub-
Mediterranean Province during the Toarcian, as a conse-
quence of the environmental structure of the two provinces.

Adaptation to an r-selective environmental regime is pos-
sible only for a small part of the populations previously
adapted to a K-selective regime, due to several factors lim-
iting the capability to carry out a r-selective strategy. Fur-
thermore, the selective pressure on the populations of a K-
selective regime constantly tends to force the populations
themselves to migrate into regions that are undersaturated
and where competition is low. Both of these contrasting fac-
tors probably affected the mediterranean ammonites.

50
submediterranean
province
levesquei
thouarsense
<
=
iz
<
5 aaawes < =
variabilis © ra
=
= a
migration of
Phymatoceras
bifrons <
migration of
Chartronia
falciferum
FIGURE 2

in Mediterranean Province and migration to Sub-Mediterranean Province

Consequently several isolated forms or small populations
of Mediterranean ammonites, belonging to chronologically
different species, ventured in successive stages and in dif-
ferent periods into the Sub-Mediterranean Province, forming
a discontinuous fossil record.

Such populations only rarely succeeded in establishing
themselves in the new selective regime and originating sub-
mediterranean phyletic lineages. We saw that in Phymato-
ceratinae this happened only twice, with the development of

GIOVANNI PINNA

mediterranean
province

meneghinii

)
<
jeg
ud
1S)
(e}
j
a
2
Re
<
oO

P| ‘| erbaense

PHYMATOCERAS

bifrons

serpentinus

HILDAITES

Phylogeny, biogeography and stratigraphic distribution of Toarcian ammonites of the subfamily Phymatoceratinae. Note origin

the phyletic lineages of the genera Haugia and Denckmannia,
derived from migrations of mediterranean populations of the
genera Chartronia and Phymatoceras.

This sequence of events produced a discontinuous fossil
record of the three mediterranean phyletic lineages in the
Sub-Mediterranean Province, a fossil record constituted both
by the transient isolated forms that repeatedly entered the
province and by the migrations that at two separate times
established permanent populations of Phymatoceras and of

FOSSIL RECORD DISCONTINUITY

Chartronia. On the contrary, in the Mediterranean Province
the fossil record of the three endemic lineages (the genera
Phymatoceras, Chartronia and Lillia) shows a continuous
and gradual evolution, and for every phyletic lineage it is
possible to identify one single form for each successive time
interval.

On the other hand, in the Sub-Mediterranean Province the
fossil record of the two endemic phyletic lineages shows a
gradual and continuous evolution, completely analogous to
the evolution observed in the Mediterranean Province within
the three phyletic lineages that are endemic to this province.

Because of the forced direction of the populating move-
ments, the fossil record of the sub-mediterranean endemic
groups is almost non-existent in the Mediterranean Province.

In conclusion, it may be inferred that in the subfamily
Phymatoceratinae groups endemic to a certain area followed
a continuous and gradual evolution, even though evolution
seems discontinuous when observed in areas that are different
from those where the groups are endemic. Because such a
scenario seems to apply to other groups of toarcian ammon-
ites (Pinna and Levi-Setti, 1971), we can generalize by stat-
ing that the discontinuity observed in many ammonite
phyletic lineages is at least partly due to biogeographic
causes, that is by movements into areas of low-selective re-
gime by populations of groups endemic to high-selective re-
gime areas.

Modality of Appearance of the Features in
Phymatoceratinae

The great abundance of individuals on the one hand, and
the planispiral shape of the shell, allowing us to follow the
ontogenetic development of the single individuals, on the
other make ammonites excellent material for the study of
both mechanisms of appearance of new characters and het-
erochronies of development. Ammonites have been used to
this aim by many palaeontologists (Waagen, Wurtemberger,
Neumayr, Hyatt, Pavlov, Schindewolf and more recently Cal-
lomon, Dommergues and Landman), who created several
evolutionary models about the relationships between onto-
geny and phylogeny.

Historically, the relationship between ontogeny and phy-
logeny in ammonites led to two opposite interpretations of
the processes by which features appear and develop.

At first, palaeontologists, including Hyatt, Buckman, True-
man and Spath (in early papers), supported a palingenetic
process for evolution in which new stages were added at the
end of ontogenesis and adult features of the ancestor were
pushed back into the juvenile stages of the descendant. Be-
cause such evolutionary process would bring about a direct
recapitulation in the history of the individual of the history
of the phyletic lineage, it always had to involve an acceler-
ated ontogenesis. Moreover, such process could be easily
explained by invoking the inheritability of the acquired fea-
tures. The main consequences of the recapitulationist vision
of evolution were the finalization of the evolutionary process
(for instance Hyatt’s theory of racial old age), orthogenetic

evolution, and the parallel evolution of phyletic lineages, re-
sulting in polyphyletism. From the standpoint of systematics,
another consequence, stressed by Donovan (1973) and re-
lated to the concept of evolution by parallel lineages, was
the huge multiplication of systematic entities, something
which has always been sought by stratigraphers.

Then, starting mainly from Pavlov’s work (1901), most
students of ammonites — among them Spath, Brinkmann and
Schindewolf — supported a cenogenetic process for evolu-
tion, in which juvenile features of the ancestors would be
shifted to the adult stages of the descendant (proterogenesis
of Schindewolf, 1936). Such an evolutionary process led to
the denial of recapitulation and acceleration, and it made
possible a “rejuvenation” of phyletic lineages (the “escape
from specialization” of Hardy, 1954). By excluding extine-
tion as the inevitable fate of every evolutionary lineage, it
justified the elimination of any finalist interpretation and,
therefore, of any intrinsic evolutionary force. Moreover, the
cenogenetic model allowed palaeontologists to overcome
some of palaeontology’s typical biases, including the direc-
tionality and the linearity of phyletic lineages as the result
of an intrinsic evolutionary power, and the idea of the pro-
gressive decrease in the evolutionary potential. And, above
all, it allowed us to suppose a genetic continuity compared
to a morphologial discontinuity.

A new interest in the role played by heterochrony in the
evolutionary processes arose in 1977, with the publication
of Gould’s volume Ontogeny and Phylogeny and with the
further analysis of different heterochronic processes by Al-
berch et al. (1979) (see McKinney 1988, McNamara 1990,
McKinney and McNamara 1991). It has been demostrated
that evolution of ammonites follows heterochronic processes
that are both cenogenetic (paedomorphosis) and palingenetic
(peramorphosis) (Landman 1988, Dommergues 1990).

In the three mediterranean phyletic lineages of the sub-
family Phymatoceratinae, evolution is expressed primarly in
the ornamentation, which becomes stronger and more elabo-
rate (in Phymatoceras, Lillia and Chartronia), and to a lesser
extent the shape of the shell resulting in increasing of the
involution (in Chartronia), and an attenuation of the ventral
furrows (in Lillia). All the modifications take place through
peramorphosis: during the evolution of the three phyletic
lineages the morphologic changes always take place with
gradual progression at the end of the ontogenetic growth, so
that the three phyletic lineages can be considered as pera-
morphoclines according to McNamara (1982). In the shell
of the most evolved forms of each phyletic lineage, up to
three different morphologic stages in a recapitulative se-
quence can be observed.

The same heterochronic model is applicable to the evolu-
tion of the sub-mediterranean genus Haugia.

Peramorphic modifications are not the general rule for the
ammonites of the mediterranean Toarcian: in the family Dac-
tylioceratidae (Pinna and Levi-Setti, 1971), for instance, the
evolution in most phyletic lineages results from paedomor-
phosis, while peramorphic processes are highly limited.

in
i)

Continuity and Discontinuity in Evolution

Whether the morphologic modifications occur through
peramorphosis or paedomorphosis, analysis of the ontoge-
netic changes in ammonites allows one to reconstruct the
phyletic lineages. What makes the analysis possible is the
fact that the spiral shape of the shell allows observation of
successive, different growth stages in the individuals (that 1s
to say the period of persistence of a given morphologic stage)
and the development of such stages during evolution (that
is to say the change in the extension of a given morphologic
stage). We observed that during evolution a new feature may
move forward starting from the early stage of ontogenesis,
and involving more and more adult stages in the case of the
paedomorphic processes: or it may move backward toward
more and more juvenile stages in the case of peramorphic
processes. When we have a very complete and therefore con-
tinuous series of forms, such shifts appear highly gradual.

The greater the completeness of the fossil record, the
greater the evidence that change has been gradual. This
means that this kind of process can easily seem to be dis-
continuous, because the improbability of demonstrating ab-
solute continuity where all generations are documented. Even
in this theoretical case the process would still seem discon-
tinuous, because the appearance of a new feature or a whole
set of features will always represent a discontinuity if com-
pared to the previous situation.

At this point it is necessary to agree on the terms “conti-
nuity” and “discontinuity” with respect to the evolutionary
process. So far as ammonites are concerned, we can say that
the successive evolutionary changes described in different
phyletic lineages by different authors consist of changes of
a comparable magnitude; the changes in the shape of the
shell or of some of its features, the appearance of some fea-
tures such as as grooves or carinae, or the changes in the
ornamentation follow one another from one form to the fol-
lowing form in the same fashion, although not at the same
rate. Despite this, such changes have sometimes been con-
sidered as discontinuous (for example, this idea was implied
by Spath) and sometimes as continuous (see Landman, 1988).
I agree with Simpson, who wrote (1944, p. 50)

‘

In its crudest form, the distinction between continuity
and discontinuity in evolution is almost meaningless.
The developed organism is absolutely discontinuous
from its parent, and any real difference between the
two, however small, is discontinuous; there is no mor-
phological continuum.

Continuity is a sequence of discontinuities, the intervals
of which can be made shorter and shorter by the progressive
improvement of the fossil record, but which can not be elimi-
nated. In fact, even if were we able to find as fossils all
organisms of all the generations that made up a given phyletic
lineage, there would still be among these organisms the dis-
continuities that are inherent to the essence of individuality.

Therefore we may paraphrase the statement of Otto Schin-
dewolf, who was one of the biggest supporters of disconti-
nuity, as follows:

GIOVANNI PINNA

The search for a series of successive transitory stages
between two types, proving the gradual creation of a
new structural type, will be vain, since the perfect con-
tinuity between two successive types does not exist.

Conclusion

As pointed out by Dommergues (1990, p. 162)

-ammonoids constitute a rather homogeneous group
organized around a single Bauplan from which only
a few groups deviate appreciably. . . . Moreover, it 1s
probable that, throughout the range of the order, most
ammonoids inhabited homologous or at least similar
marine ecosystems. These morphofunctional and eco-
logical continuities allow extrapolation of general pa-
laeobiological assumption from only a few detailed
analyses of trends... .

Dommergues statement justifies extrapolating the results
of the study of the mediterranean and sub-mediterranean
Phymtoceratinae, and concluding that the biggest disconti-
nuities found in the evolution of phyletic lineages among
ammonites in general cannot be ascribed to discontinuities
in the evolutionary process. Major discontinuities are not of
phyletic origin, but are due to biogeographic factors, while
lesser ones can be ascribed to the discontinuity that is in-
trinsic to individuality itself.

Literature Cited

Alberch, P., S.J. Gould, G. F. Oster, and D. B. Wake. 1979. Size
and shape in ontogeny and phylogeny. Paleobiology 5:296-317.

Arkell, W. J.. B. Kummel, and C. W. Wright. 1957. Mesozoic Am-
monoidea. Pages 80-465 in R.C. Moore, editor. Treatise on In-
vertebrate Paleontology, L: Mollusca. Volume 4. Geological So-
ciety of America, Lawrence

Dommergues, Jean-L. 1990, Ammonoids. Pages 162-187 in K. J.
McNamara, editor. Evolutionary Trends. University of Arizona
Press, Tucson,

Dommergues, Jean-L., B. David, and D. Marchand. 1986. Les re-
lations ontogenese-phylogenese: applications paleontologiques.
Geobios 19:335—356.

Donovan, D. T. 1973. The influence of theoretical ideas on am-
monite classification from Hyatt to Trueman. Contributions in
Paleontology University of Kansas 62:1—16.

Gabilly, J. 1976, Evolution et systematique des Phymoceratinae et
des Gammoceratinae (Hildocerataceae: Ammonitina) de la region
de Thouars, stratotype du Toarcien. Memoires de la Societe
Geologique de la France 124:1—90.

Gould, S.J. 1977. Ontogeny and Phylogeny. Harvard University
Press, Cambridge

FOSSIL RECORD DISCONTINUITY

Hallam, A. 1975. Jurassic Environments. Cambridge University
Press, Cambridge.

Hardy, A. C. 1954. Escape from specialization. Pages 122-142. in
J. Huxley, A.C. Hardy and E. B. Ford, editors. Evolution as a
Process. George Allen and Unwin, London.

Landman, N. H. 1988. Heterochrony in ammonites. Pages 159-182
in M.L. McKinney, editor. Heterochrony in Evolution: a Mul-
tidisciplinary Approach. Plenum Press, New York.

McKinney, M. J. 1988. Heterochrony in Evolution: a Multidisciphi-
nary Approach. Plenum Press, New York.

McKinney, M. J., and K.J. McNamara. 1991. Heterochrony: the
Evolution of Ontogeny. Plenum Press, New York.

McNamara, K. J. 1982. Heterochrony and phylogenetic trends. Pa-
leobiology 8:130—142.

. 1986. A guide to the nomenclature of heterochrony. Journal
of Paleontology 60:4—13.

McNamara, K. J., editor. 1990. Evolutionary Trends. University of
Arizona Press, Tucson.

Pavlov, A. P. 1981. Le Crétace inférieur de la Russie et sa faune.

nN
oe)

Mémoires de la Societe Imperiale des Naturalistes de Moscow
16: 1-87.

Pinna, G., and F. Levi-Setti. 1971. I Dactylioceratidae della Provin-
cia Mediterranea (Cephalopoda Ammonoidea). Memorie della
Societa Italiana di Scienze Naturali e del Museo Civico di Storia
Naturale di Milano 19:49—136.

. 1973. Note su uno studio delle ammoniti liassiche della
sottofamiglia Phymatoceratinae Hyatt, 1900. Bollettino della So-
cieta Paleontologica Italiana 12:130—142.

Sanders, H. L. 1968. Marine benthic diversity: a comparative study.
American Naturalist 102:243—282.

. 1969. Benthic marine diversity and the stability-time hy-
pothesis. Brookhaven Symposia in Biology 22:71—81.

Schindewolf, O. H. 1936. Palaontologie, Entwicklungslehre und Ge-
netik: Kritik und Synthese. Gebriider Borntraeger, Berlin.

Simpson, G. G. 1944. Tempo and Mode in Evolution, Columbia
University Press, New York,

Smith, A. G., D.G. Smith, and B. M. Funnell. 1994. Atlas of Meso-
zoic and Cenozoic Coastlines. Cambridge University Press, Cam-
bridge.

Ww
wn

SEGMENTS, BODY REGIONS, AND THE CONTROL OF DEVELOPMENT THROUGH TIME

Alessandro Minelli

Department of Biology
University of Padova
Via Trieste 75
I 35121 Padova, Italy

Four main dimensions of structural and/or developmental complexity are identified in the organization and life history of
arthropods. These dimensions correspond to the antero-posterior and dorso-ventral body axis, to the proximo-distal axis of
the appendages and to the temporal axis of (post-embryonic) development. Patterns of change are very often concordant
along two or more of these dimensions. No structural or developmental axis is partitioned into more than about 5 major
units (e.g., tagmata). Speculations are offered as to the possible relationships between these aspects of organizational complexity
and Kauffman‘s (1993) model of genomic organization. The importance of the segmental organization of the body as a basic
structural condition is de-emphasized. Tagmata, rather than segments, are regarded as the primary target of natural selection.

Segmented Animals

Early in the nineteenth century, Cuvier (1817) pointed to
segmentation as a basic feature of animal morphology. Seg-
mentation was characteristic of the Articulata, one of his
four main divisions (embranchements) of the animal king-
dom. Cuvier‘s Articulata embraced the segmented groups
that we currently identify as the annelids and the arthropods,
but not the vertebrates, whose organisation Cuvier regarded
as different enough from that of the other segmentally or-
ganised animals as to warrant the status of an independent
embranchement. However, even the seemingly unlikely com-
parison between vertebrate and arthropod segmentation was
proposed by Etienne Geoffroy Saint-Hilaire, Cuvier's friend-
rival at the Muséum d‘Histoire Naturelle in Paris in one of
his most hazardous comparisons. Geoffroy’s (1820a, 1820b,
1822) efforts to treat an arthropod as a kind of vertebrate
living inside its vertebrae was all too easily ridiculed,
whereas the homology between annelid and arthropod seg-
mental organisation has remained unchallenged until very
recently.

Today, we approach the study of segmentation from the
vantage point of developmental biology and especially of
the molecular genetics of development. The new evidence
definitively confirms the very perceptive point of view of
the Russian comparative morphologist Beklemishev (1969
[1964]), who interpreted segmentation as a kind of symmetry
which is not a uniquely derived, hence homologous, feature.
Rather, he considered that segmentation easily emerges as a
formal concept of rational morphology, on the one hand, and
as a concept of functional anatomy, on the other. There is
no longer any doubt, that segmentation develops in different
animal groups through different morphogenetic pathways.

In insects, segmentation goes on, basically, within the (pro-
spective) ectoderm, whereas in vertebrates and annelids seg-
mentation starts within the mesoderm. Cell proliferation may,
or may not, be required for segmentation. In some animals
the segments appear synchronously, as in Drosophila; in oth-
ers, and more commonly, in an antero-posterior sequence.
New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

In my view, there is now evidence enough to regard annelid
and arthropod segmentation, not to say vertebrate segmen-
tation, as independently derived. If this view will prove to
be true, then the major support to Cuvier’s Articulata will
be seriously shaken. In this context it may be useful to point
to recent molecular evidence suggesting a not-so-close rela-
tionship between arthropods and annelids (Field et al., 1988;
Eernisse et al., 1992; Ballard et al., 1992; Whitington, 1993).

In this paper, I will discuss some hitherto overlooked as-
pects of the body organisation in one of these groups of
segmented animals: the arthropods.

Four Dimensions of Arthropod Complexity

McShea (1993) has offered a comparative quantitative
analysis of the structural complexity of the vertebral column
of mammals. His paper and, in particular, the complexity
indices introduced therein, provide a valuable starting point
for this kind of study. However, in the present analysis |
will follow a less formal approach, simply applying the fol-
lowing intuitive principles:

1. complexity is not dependent on the number of parts:
strictly homonomous sequences of 10, 20 and 100 segments
are, as such, of equal complexity;

2. complexity is increased, however, when changing from
an unsegmented to a segmented condition;

3. complexity is only slightly increased when changing
from a strictly homonomous pattern to a progressive, gradi-
ent-like pattern;

4. complexity is highly increased, when developing sharp
structural boundaries, as between head and trunk in an in-
sect’s body;

5. the above principles may also apply to the time dimen-
sion, if developmental stages are equated to body segments
and major developmental phases (such as the insect larva,
pupa, and imago) are equated to body regions.

To approach a complexity analysis of arthropod organisa-
tion, it seems advisable to identify four distinct although re-
lated dimensions of the “segmented” architecture of these

animals. These four dimensions (Fig. |) are the antero-pos-
terior and the dorso-ventral body axis, the proximo-distal
axis of the body appendages (such as legs or antennae) and
the temporal axis of the individual developmental time.

Generally speaking, the longitudinal body axis of arthro-
pods is not at all like the uniform sequence of identical seg-
ments that we see in many annelids. The body of arthropods
is divided into regions (tagmata), such as the head, thorax
and abdomen of insects and the prosoma and opisthosoma
of spiders. The number of tagmata is always very small and
quite independent of the number of body segments. For in-
stance, the longitudinal complexity (tagmosis) of centipedes
is more or less the same (possibly, even a bit higher) in
lithobiomorphs, with 15 leg-bearing segments in the adults,
as in geophilomorphs, with up to 191 leg-bearing segments.
The same pattern occurs within millipedes, or branchiopod
crustaceans.

More often than not, the actual complexity of the longi-
tudinal axis of an arthropod is sensibly larger than suggested
by the conventional, admittedly stronger, partitioning into
two to four tagmata. For instance, the second body region
of malacostracan crustaceans (pereion), usually consists of
an anterior section, whose appendages are developed as max-
illipedes, and a posterior section, whose appendages retain
the more conventional shape of locomotory limbs. However,
within any major region of the body, segments and append-
ages generally retain a distinctive uniformity or, commonly,
change in a continuous, gradient-like manner. These changes
could be obtained, in principle at least, with a very modest
amount of information, in addition to that required for build-
ing one segment, and for uniformly repeating this segmental
unit many times. Therefore, these gradual changes do not
sensibly increase the complexity of body architecture. Things
are different at the transition points between one region and
another, where virtually everything changes, in the external
aspect as well as in the internal anatomy. We cannot avoid
getting the impression that a lot of different genes are se-
lectively expressed in the individual regions. | am reminded
of a notion of plant cell biology: an average plant cell might
have something like 20,000 distinct MRNA sequences, 1000
of which, perhaps, are different from those found in another
cell type within the same plant (Kamalay and Goldberg,
1980; Alberts et al., 1983). Certainly, this 1s a point badly
in need of extensive comparative investigation.

Moving to the second dimension of complexity, 1.e., to
the dorsoventral body axis, It is easy to provide examples
of dorsal and ventral aspects of the same animal behaving
quite independently from one another, even in respect to
gross segmentation. A strong dorso-ventral partitioning into
two distinct domains 1s evident, for instance, in pauropods
and in symphylids. In several instances, however, the dor-
soventral axis seems to allow a partition into more than two
domains. A good example is the thorax of the insects. Here,
the levels where wings and legs respectively articulate with
the segment are quite obvious limits between tergal and pleu-
ral, and pleural and sternal partitions.

The third axis of complexity is that of the appendages. In
this case too, we can have a very diverse, and often quite

ALESSANDRO MINELLI

high, number of segments, but these are never structured
into a very high number of “regions.” The very high number
of flagellomeres forming the terminal section of the legs in
the house centipede (Scutigera) or in many harvestmen, does
not make the legs of these animals more complex relative
to the legs of other centipedes and arachnids. Meristic vari-
ation does occur at all taxonomic scales, without seemingly
altering the overall complexity of the appendage. For in-
stance, the number of tarsal joints is generally stable in
Coleoptera; but the rule of having 5 of them has widespread
exceptions, sometimes scattered, sometimes clustered within
one family. The same holds true for the number of anten-
nomeres. Most beetles have 11 of them, but this number,
very seldom exceeded, is quite often reduced to 10, 9 or
even less, without sensibly affecting the shape, or the com-
plexity, of the appendage. The above remark about morpho-
logical changes in a gradient along the main axis of the body
applies to the axis of appendages as well. The apical clubs,
so common in the antennae of beetles and other insects, e.g.,
Heteroptera, Orthoptera and even Lepidoptera, are examples
of easy modifications of a basically homonomous appendage.

To these three morphological axes we can now compare
a fourth, apparently unrelated, axis: that of developmental
time. At first sight, any similarity between the structure, or
complexity, of developmental schedules on the one hand,
and the structure, or complexity, of the longitudinal, dor-
soventral and appendicular axes of the body on the other,
could be suspiciously regarded as merely coincidental. How-
ever, a causal link between the temporal dimension of de-
velopment and the spatial dimension of the longitudinal body
axis has been definitely demonstrated by the molecular ge-
netics of development. A detailed, causal relationship has
been ascertained between the temporal sequence of transcrip-
tion and expression of individual Hom/Hox genes, in Droso-
phila as well as in vertebrates, including the specification of
the individual spatial domains under the control of these
genes, along the antero-posterior axis of the body. In a few
words: earlier equals anterior, later equals posterior. This
equivalence has nothing to do with segmentation per se.
Rather, it is probably a much more general feature of meta-
zoan organisation, the core of Slack‘s et al. (1993) concept
of zootype. Moreover, there is a hard-wired feature, behind
this remarkable correspondence between developmental time
and positional specifications along the body axis, 1.e., the
much conserved linear arrangement of these homeotic genes
along the chromosomes, a linear arrangement which is se-
quentially expressed, thus giving rise to a sequence of body
structures that is collinear with the gene sequence.

In post-embryonic development there are two kinds of tem-
poral units comparable to the morphological units that we
call segments and tagmata. The temporal “segments” of de-
velopment are the individual ontogenetic stages separated by
a moulting event. Most moults are followed by minor
changes only, hardly affecting anything besides body size
and variations of colour and cuticular patterns, including the
number of setae on selected body parts. In addition, there
may be an increase in the number of body segments (in
anamorphic arthropods, such as millipedes), but this is a sim-

SEGMENTS, BODY REGIONS AND DEVELOPMENT

Nn
—

FIGURE |. The four dimensions of arthropod complexity discussed in the text. Large arrow: temporal axis of developmental time. Small
arow, left: proximo-distal axis of the appendage. Small arrow, center: antero-posterior body axis. Small arrow, right: dorso-ventral body axis

ple meristic change not entailing any sensible increase in
complexity. The changes accompanying all these moults are
the developmental equivalent of the smooth, gradual changes
in morphology that we can observe within one tagma along
the antero-posterior axis of the body. A few moults, however,
are accompanied by major changes deserving the name of
metamorphosis. These major developmental events are al-
ways few in number, as are those separating larva from pupa
and pupa from adult in holometabolous insects.

Correlations

A striking feature of arthropod complexity is that within
the same animal degrees of, or variations in, complexity
along any one of these four axes are very often coupled,
with equivalent degrees of, or variations in, complexity along
one, or more, of the other axes.

Examples of a correspondence between the complexity of
the antero-posterior axis and the complexity of developmen-
tal schedules are found in millipedes and hymenopterans.

All millipedes (Diplopoda) develop by anamorphosis, 1.e.,
they hatch from the egg as larvae with a small number of
segments and progressively add more and more segments
moult after moult. There are, however, three different kinds
of millipede anamorphosis, recently redefined by Enghoff et
al. (1994) as euanamorphosis, teloanamorphosis and hemi-
anamorphosis, exemplified, respectively, by julid, chor-deu-

matid and glomerid millipedes. In euanamorphosis, the num-
ber of moults is not strictly defined nor is the number of
segments, which continuously increases through life. In
teloanamorphosis, the number of moults is fixed, as is the
number of segments, which are progressively added at each
moult. In hemianamorphosis, the number of segments is
fixed, but it is reached after a few moults: afterward the
animal continues to moult, but without any further increase
in the number of segments. The complexity of these devel-
opmental schedules increases from euanamorphosis through
teloanamorphosis to hemianamorphosis. Correspondingly,
the complexity in the regionalisation of the body increases,
from the very uniform trunk of the euanamorphic julids to
the much more specialised trunk of hemianamorphic
glomerids, with gross specialisations affecting the foremost
and the hindmost segments of the body.

In Hymenoptera, we can contrast the plesiomorphic devel-
opment of sawflies (with several, morphologically uniform,
caterpillar-like larval stages), with the highly derived devel-
opment of many Parasitica, whose developmental schedule
is more complex. This complexity manifests itself with a
first larva, often of bizarre unsegmented shape, that 1s mark-
edly different from the second, or mature, larva, which is of
more conventional shape. Again, as in millipedes, this con-
trast is mirrored by the contrast between the simple, plesio-
morphic abdomen of sawflies and the highly modified, and
sometimes very complex, abdomen of many Parasitica. What

FIGURE 2. Antenna of the male of the blister beetle Me/oe proscara-
baeus (Coleoptera: Meloidae)

is Interesting for our argument 1s that a more complex pattern
also evolves here along the dorsoventral axis. Lateral to the
main tergal plates, a couple of laterotergites are split off on
several abdomen segments of representatives of a number
of families of these hymenopterans, e.g., segments 3—7 in
Platygasteridae and segments 2—6 in Scelionidae.

In other instances, an unusually complex post-embryonic
development 1s accompanied by unusually complex append-
ages.

One example involves the blister beetles (Meloidae), a
family of beetles with hypermetamorphic post-embryonic de-
velopment. The standard developmental schedule of blister
beetles comprises a primary larva (the very mobile triun-
eulin), followed by a secondary larva of very different shape,
which in turn is followed by a resting pupa-like stage and,
again, by an additional active larva, more or less similar to
the secondary one, and finally by the conventional pupa and
the adult! With this unusual developmental complexity we
can match the morphological complexity of the antennae of
many (not all!) blister beetles. This feature deserves particu-
lar attention, because it is not simply an example of that
“lesser” structuring we find in the antennae of many beetles,
with more or less distinctly gradient-like changes in the shape
of the basal or the terminal antennomeres. Of much more
interest to us, the modifications we see in the antennae of a
few meloid beetles do not affect the proximal or the distal
antennomeres, but exclusively a few intermediate ones!
(Fig. 2) That means that blister beetles can operate a shape
control in a region of the appendage that usually exhibits no
plasticity at all, Thus, in blister beetles antennal complexity
parallels developmental complexity, by obtaining a control

ALESSANDRO MINELLI

over the mid-point of a previously uniform sequence (of an-
tennomeres, or of developmental stages). I do not think that
the unique antenna of Me/oe occurs in a hypermetamorphic
beetle merely by chance.

More examples are to be found among arachnids. Here,
the only groups that undergo sensible post-embryonic devel-
opmental changes, with clearly distinct larval/nymphal stages
(true metamorphosis), are mites and ricinulei. These arach-
nids go through a six-legged larval stage, before getting the
eight-legged nymphal and adult condition. However, this de-
velopmental complexity is not without a counterpart in the
complexity of the main body axis, as is suggested by the
cumbersome and idiosyncratic nomenclature introduced to
identify the regions we can often distinguish within the body
of a mite: gnathosoma, hysterosoma, idiosoma, meta-po-
dosoma, opisthosoma, propodosoma, podosoma. These terms
are more numerous than those used for spiders or scorpions.
In mites, this complexity has no clear counterpart in the struc-
ture of the legs, but this fact 1s possibly explained by the
very small size of these arthropods, a formidable limit to
any further attainment of complexity beyond the standard
degree of arachnid legs. At least there is no place, in the
legs of mites, for more patterned articulations. On the other
hand, very complex apical structures, in the shape of adhesive
pads, hooks, ete., are common in mites. This explanation ts
supported by the slightly complex leg of Ricinulei, which
are not so small as mites. More generally, the effect of an
extremely small body size (or cell number?) on the devel-
opment, or the maintenance, of structural complexity 1s evi-
dent in several arthropods, a good example being the fading
of segmentation in the gall-building eriophyid mites, whose
total length 1s often less than 100 um. In these cases of size-
induced reduction, the indiscernibility of parts may open for-
midable problems of homology (Wagner 1989).

In several instances, a striking correspondence 1s found
between the structural complexity of the main body axis and
the structural complexity of the appendages. The two groups
of arachnids in which the body ends with a multi-segmented,
flagellum-like section (Palpigradi and Uropygi) also have
strikingly segmented tarsi, especially on the first pair of legs.
Again, the more evidently segmented body of Amblypygi,
in comparison to their sister group (Araneae, the spiders), is
matched by the uniquely high degrees of segmentation of
the tarsi of the | leg pair in Amblypygi, without any coun-
terpart in spiders.

Bristletails (Archaeognatha) and silverfishes (Zygaentoma)
are the hexapod equivalent of Palpigradi and Uropygi in that
their body ends in a multi-segmented filum terminale, whose
flagellar organisation is matched by that of the cerei and
other appendages.

In other instances, the correspondence between longitudi-
nal body axis and appendicular axis is found in their respec-
tive developmental schedules. One example is provided by
centipedes. Within this group some forms (Scutigeromorpha
and Lithobiomorpha) develop by anamorphosis, 1.¢., by pro-
gressive increase in the number of segments during post-em-
bryonic development. In these centipedes, the number of an-
tennal articles also increases during post-embryonie life. In

SEGMENTS, BODY REGIONS AND DEVELOPMENT

contrast, those centipedes (Scolopendromorpha and Geo-phi-
lomorpha) which develop epimorphically already having
their total segment complement when hatching. These cen-
tipedes also have fully-formed antennae at birth, with 14
articles in all Geophilomorpha and 17 in most Scolopen-dro-
morpha.

A more impressive correspondence between antero-poste-
rior axis of the body and proximo-distal axis of the append-
ages is offered by copepods. As recently pointed out by
Izawa (1991), Hulsemann (1991) and Ferrari (1993), during
their copepodid stages these tiny crustaceans add new seg-
ments to their appendages, according to the same “progres-
sion rule” that they follow in adding new segments to the
body. Copepods differentiate one more body (or appendage)
segment after each moult, always in subdistal position, until
the final pattern is achieved. However, that is not yet the
whole story. The body is characteristically marked by a me-
chanically important flexure, which may be placed in front
of the fifth pair of legs or just behind them (thus allowing
the traditional distinction between podoplean and gym-
noplean copepods). This major singularity, which is clearly
superimposed to the basic tagmosis (as demonstrated by its
ontogenetic shifting from a more anterior position to its final,
podoplean or gymnoplean position) is mirrored in the ap-
pendages by the geniculation of the antennules of most male
copepods, a segmental singularity of the same kind as that
of the antenna of Meloe just referred to.

Other examples clearly show a threefold/fourfold corre-
spondence in structural complexity between body, appendage
and development.

A first example concerns branchiopod crustaceans. In this
group, there is a large range of structural models and devel-
opmental schedules, whose complexity changes in a con-
certed way. In the Cambrian Rehhachiella (Walossek, 1993),
the number of segments is fairly high, as is the number of
podomeres forming each of the body appendages, as is the
number of moults undergone by the animal during its post-
embryonic life. In addition, the body segments are very uni-
form, except for the transition between leg-bearing and
legless segments; the appendicular joints are also very uni-
form; and the sequence of stages during ontogeny is very
regular, indeed, at the highest degree we currently know for
any arthropod. Specialisation, 1.e., the attainment of com-
plexity, increases in anostracans, conchostracans and ostra-
cods, in the order, but always bringing together correspond-
ing chan-ges along all three main dimensions (body,
appendage, developmental time).

Additional lines of evidence support the hypothesis that
all of the main dimensions of arthropod organisation evolve
in a concerted way. One of these is the occurrence of “pro-
gradous” development. By this | mean a kind of exception
to the rule that morphogenesis proceeds in antero-posterior
and proximo-distal sequence. Formally equivalent, in a sense,
to these progradous disto-proximal and postero-anterior se-
quences of morphogenesis are such events as programmed
cell death by apoptosis, segment fusion and the like. All of
these events are indeed rare and all of them surely require
additional genetic information and strong developmental con-

trol to overcome the nearly universal retrogradous trend. It
is worth noting here that true segment fusion is all but com-
mon. What is generally referred to under this name is non-
disjunction, 1.e., the lack, rather than the reversal, of a basic
developmental process. A few examples of progradous de-
velopment are given in the following three paragraphs.

As to the direction of the differentiation wave along the
body axis, two good examples are found in the stomatopods
and in the copepods. Amongst stomatopods, there are species
in which the trunk appendages regularly differentiate in an-
tero-posterior sequence, while in others the appendages of
the hindmost region of the body (pleopods), develop before
the appendages of the mid-body region (pereiopods). In co-
pepods, similar instances of the anticipated maturation of
some pair of appendages before those in front of them have
been recently illustrated by Ferrari (1993).

Something like progradous development in the morpho-
genesis of appendages has been described for the antennae
of a large moth, Antheraea polyphemus, by Steiner and Keil
(1993). During the pupal stage, the antennae are initially
developed as leaf-like blades. These then become progres-
sively segmented through the effect of incisures developing
from the edge of the blade towards the prospective rhachis
of the antenna.

Much more intriguing is the case of the so-called gonopods
of male helminthomorph millipedes. In these animals, with
the moult leading to the first (or only) mature stadium, one
or two pairs of legs, normally developed as walking legs in
the previous stages, are abruptly changed into sexual ap-
pendages of unusual, unsegmented, and often extremely com-
plex shape. Interestingly, this specialisation is sometimes re-
versible. In a few julid species, sexually mature males may
moult into so-called intercalary males, which are not sexually
mature and bear “sexual” appendages of a shape somewhat
midway between typical walking legs and full formed
gonopods. A further moult, again leading to a mature con-
dition with typical gonopods, is occasionally observed.
Somewhat parallel events are known in peracarid crusta-
ceans, especially tanaidaceans.

Arthropod complexity: some speculations

It is tempting to speculate about the common background
to all these dimensions and variations of complexity. Articu-
lating the body into regions means to fix a small number of
strong markings along the main axis of the body. These mark-
ings act as “hot spots” (Minelli and Schram, 1994) for the
subsequent expression of whole alternative networks of struc-
tural genes. The same seems to go on with partitioning the
appendages into a few major segments. The same also seems
to go on, although perhaps to a lesser degree, with the dis-
tinction of two or three domains along the dorsoventral axis
of the body. The same. finally, goes on with partitioning the
developmental sequence into a few major temporal segments,
at least in the case of strongly patterned metamorphoses, as
in holometabolous insects and many crustacean groups.

This multidimensional correspondence between structuring

processes 1s, perhaps, the expression of related underlying
mechanisms.

Recent experimental work on Drosophila (reviewed in An-
derson, 1995) demonstrates that one and the same molecular
signal, the product of the gene Gurken, instructs the follicular
cells in the ovary to impose both the antero-posterior and
the dorso-ventral polarity upon the developing oocyte.

On the other hand, there are many indications, in insects,
of extensive similarities and even overlappings in the mecha-
nisms of genetic control of segmentation and specification
of regions along the longitudinal body axis and the appen-
dicular one (e.g., Carroll, 1994).

Along each one of the time/space dimensions we are dis-
cussing here, arthropods become partitioned into a small
number of units, the transitions (e.g., from thorax to abdo-
men, or from pupa to adult) being marked by a more or less
dramatic choice between alternative developmental subrou-

tines, possibly involving a number of genes which are
switched off in the neighbouring domains.

As already remarked, there seems to be a /ow upper limit
to the number of body regions. | wonder whether regions
could be regarded as a kind of units in competition for the
access to some resources, much in the sense of the compe-
tition between alternatively specialised cell-lineages within
one individual animal, according to the model developed by
Leo Buss (1987) in his book The Evolution of Individuality.
More recently, Buss (1990) has suggested that body segments
might be regarded as competitive units, but I think that such
competition, if it exists at all, must be very slight. Indeed,
segments are in principle equivalent units, like the cells of
the same clone. They obviously are in competition for ma-
terials, but not for genetic resources, 1.e., for activation (or
de-repression) of alternative genetic networks. Tagmata, in-
stead, are basically heterogeneous units, possibly relying on
alternative genetic networks, whose number (to follow Kauff-
man, 1993) could be small and thus potentially limiting.

However, there seems to be no need to embrace wild
speculations about the functional organization of the genome.
To quote Kauffman (1993:53—54), “as systems with many
parts increase both the number of those parts and the richness
of interactions among the parts, it 1s typical that the number
of conflicting design constraints among the parts increases
rapidly. Those conflicting constraints imply that optimisation
can attain only ever poorer compromises. No matter how
strong selection may be, adaptive processes cannot climb
higher peaks than afforded by the fitness landscape. That 1s,
this limitation cannot be overcome by stronger selection.”
rhis means that the upper limit to the number of interacting
parts (here, tagmata) must always be small, even in what we
like to call higher organisms. Maybe this limit is more strin-
gent here, due to the synorganisation of integrated, co-
adapted body regions.

Problems and prospects

The concerted variation in complexity of the morphologi-
cal and developmental body axes of the arthropods is not so

ALESSANDRO MINELLI

general as the previous pages may suggest. Some contradic-
tory evidence and problematic areas shall be cited here.

A first difficulty is that many arthropods can regenerate
appendages, or parts thereof, whereas no arthropod is able
to regenerate body segments. In this respect, there seems to
be no equivalence between the longitudinal and the appen-
dicular axes.

Another problem is that, while dealing with the temporal
dimension of development and its complexity, I have delib-
erately dealt with post-embryonic development only. As for
the embryonic portion of development there are several ob-
scure areas. One of these 1s that the first developmental stages
are not yet under zygotic control; we could perhaps describe
them as a kind of extended, or prolonged, phenotype of the
mother.

Another difficult question is, whether it is meaningful to
interpret the temporal partitioning of the embryonic devel-
opment along the same lines that we use to distinguish a
larva from a pupa from an adult beetle or fly.

I have deliberately de-emphasized segments in deference
to the importance of structural and developmental complex-
ity. In a developing arthropod, segments behave as building
blocks, not necessarily more complex than are other building
blocks of developing organisms (cells, for instance, or em-
bryonic sheets). Segmented organization at some stage is
probably important to ensure regularity of patterns, including
the equal spacing of transient or definitive structures. Think
of the problems in establishing a correctly wired nerve cord
or providing for the mechanically important regularity in the
distribution of coelomic pouches and body wall muscles in
a typical annelid. No wonder, that a segmental arrangement
of equipotent cells is also transiently expressed by animals
which, later on, do not exhibit the slightest trace of segmen-
tation, such as nematodes (Salser and Kenyon, 1994).

Segmentation, however, can also have a functional value
in the fully-developed animal, especially in locomotion. This
circumstance has obviously been central in establishing the
body plan of arthropods. However, as soon as such a func-
tional constraint is no longer at work, segmentation Is re-
duced or disappears altogether. However, in this case too
the complexity of the body plan and that of its development
(a complexity originally built on the foundation of a seg-
mented germ) is not lost together with segmentation. This
complexity is a much more reliable expression of the genetic
architecture of the animal, than segmentation per se. Of
course, this is not a privilege of arthropods, but arthropods,

just because of their segmented, articulated organization, are
J g g

perhaps more suitable to rational dissection of the issues than
are other animals. In this way, the manifold morphological
and developmental complexity of arthropods may become a
paradigm for metazoans generally.

Acknowledgements
I am very grateful to Michael T. Ghiselin and Giovanni

Pinna for inviting me to the very stimulating San Francisco
workshop. In_ particular, I most sincerely acknowledge

SEGMENTS, BODY REGIONS AND DEVELOPMENT

Mike's very kind hospitality, as well as the generous support
provided by the California Academy of Sciences.

An earlier draft of this paper was carefully read by Michael
T. Ghiselin, James R. Griesemer, Nicholas D. Holland and
Frederick R. Schram: they provided me with detailed and
useful comments, for which | am very grateful.

The work has been supported, in part, by grants from the
Italian C. N.R. and M. U.R.S. T.

Literature Cited

Alberts, A., D. Bray, J. Lewis, M. Raff, K. Roberts, and J. D. Wat-
son. 1983. Molecular Biology of the Cell. New York, Garland.

Anderson, K. 1995. One signal, two body axes. Science 269:489—
490.

Ballard, J. W., G. J. Olsen, D. P. Faith, W. A. Odgers, D. M. Rowell,
and P. W. Atkinson. 1992. Evidence from 12S ribosomal RNA
sequences that onychophorans are modified arthropods. Science
258:1345—1348.

Beklemishev, W. N. 1969. Principles of Comparative Anatomy of
Invertebrates. Oliver and Boyd, London. (Vol. I Promorphology,
Vol. 2 Organology. Translation of 3rd. Ed., 1964.)

Buss, L. W. 1987. The Evolution of Individuality. Princeton Uni-
versity Press, Princeton.

. 1990. Are zooids, polyps and metameres interactors? IC-
SEB IV, Fourth International Congress of Systematic and Evo-
lutionary Biology, College Park, 1—7 July, 1990, abstracts, p.n.n.

Carroll, S. B. 1994. Developmental regulatory mechanisms in the
evolution of insect diversity. Development 1994(Supple-
ment):217—223.

Cuvier, G. 1817. Le Regne Animal Distribue d‘apres son Organi-
sation, pour Servir de Base a L*Histoire Naturelle des Animaux
et D‘Introduction a L*Anatomie Comparée, Ist Edition.
Deterville, Paris.

Eernisse, D.J., J.S. Albert, and F. E. Anderson. 1992. Annelida
and Arthropoda are not sister taxa: a phylogenetic analysis of
spiralian Metazoan morphology. Systematic Biology 41:305—330.

Enghoff, H., W. Dohle, and J. G. Blower. 1994. Anamorphosis in
millipedes (Diplopoda)— the present state of knowledge and
phylogenetic considerations. Zoological Journal of the Linnean
Society 109:103—234.

Ferrari, F. D. 1993. Exceptions to the rule of development that an-
terior is older among serially homologous segments in postmax-
illipedal legs in copepods. Journal of Crustacean Biology 31:367—
368.

Field, K. G., G. J. Olsen, D. J. Lane, S. Giovannoni, M. T. Ghiselin,
E.C. Raff, N. R. Pace, and R. A. Raff. 1988. Molecular phylo-
geny of the animal kingdom. Science 239:748—753.

Geoffroy Saint-Hilaire, E. 1820. Premiere mémoire sur le squellette

chez les Insectes, dont toutes les pieces identiques entre elles,
dans les divers ordres du systeme entomologique, correspondent
a chacun des os du squellette dans les classemens superieurs.
Memoires du Musee d*Histoire Naturelle, Paris 5:340—351.

———. 1820. Seconde memoire; sur quelques regles fond-amen-
tales en philosophie naturelle. Mémoires du Musée d*‘Histoire
Naturelle, Paris 6:31—36.

———. 1822. Considerations generales sur la vertebre. Mémoires
du Musee d‘Histoire Naturelle, Paris 9:88—119.

Hulsemann, K. 1991. The copepodid stages of Drepanopus forci-
patus Giesbrecht, with notes on the genus and a comparison to
other members of the family Clausocalanidae (Copepoda Ca-
lanoida). Helgolander wissenschaftliche Meeresunter-suchungen
45:199-224,

Izawa K. (1991) Evolutionary reduction of body segments in the
poecilostome Cyclopoida (Crustacea: Copepoda). Plankton Soci-
ety of Japan, Special Volume, pp. 71-88.

Kamalay, J. C., and R. B. Goldberg. 1980. Regulation of structural
gene expression in tobacco. Cell 19:935—946.

Kauffman, S. A. 1993. The Origin of Order: Self-Organization and
Selection in Evolution. Oxford University Press, New York

Maynard Smith, J. 1960. Continuous, quantised and modal vari-
ation. Proceedings of the Royal Society of London 152:397-409.

McShea, D. W. 1993. Evolutionary change in the morphological
complexity of the mammalian vertebral column. Evolution
47:730-740,

Minelli, A., and F. R. Schram. 1994. Owen revisited: a reappraisal
of morphology in evolutionary biology. Bijdragen tot de
Dierkunde 64:65—74.

Salser, S. J.. and C. Kenyon. 1994, Patterning in C. elegans: ho-
meotic cluster genes, cell fates and cell migrations. Trends in
Genetics 10:159—164.

Slack, J.M. W., P. W.H. Holland, and C. F. Graham. 1993. The
zootype and the phylotypic stage. Nature 361:490-492.

Steiner, C., and T. A. Keil. 1993. Morphogenesis of the antenna of
the male silkmoth, Antheraea polyphemus. IV. Segmentation
and branch formation. Tissue and Cell 25:447-464.

Wagner, G. P. 1989. The biological homology concept. Annual Re-
view of Ecology and Systematics 20:51—09,

Walossek, D. 1993. The Upper Cambrian Rehhachiella and the phy-
logeny of Branchiopoda and Crustacea. Fossils and Strata 32:1—
202.

Whitington, P. M. 1993. Conservation vs. change in early axono-
genesis in arthropod embryos—-a comparison between
myriapods, crustaceans and insects. /2 W. Kutsch and O. Breid-
bach, editors. The Nervous System of Invertebrates: A Compara-
tive Approach. Birkhauser, Basel.

HOMOLOGY, HOMEOBOX GENES, AND THE EARLY EVOLUTION
OF THE VERTEBRATES

Nicholas D. Holland

Marine Biology Research Division, Scripps Institution of Oceanography
La Jolla, California 92093-0202

During the last decade, developmental genes have been found to be remarkably conserved from one metazoan phylum to
the next. This conservation has implications for animal evolution. Among other things, developmental genetic data are now
helping to indicate homologies between body regions of distantly related animals. After briefly introducing developmental
evolution and homology, the present paper illustrates how this approach can help establish homologies between animals with
moderately divergent body plans. The focus is on work relating to the origin and early evolution of the vertebrates. A first
example uses the protein products of engrailed genes to compare mandibular arch muscles of jawless versus jawed vertebrates.
The results are relevant to the question of the origin of vertebrate jaws. A second example uses Hox genes to compare central
nervous system regions of cephalochordates versus vertebrates. The results raise the possibility that the proximate invertebrate
ancestor of the vertebrates had a relatively large brain. The success of these rather conservative applications of the method
forcefully raises the following question: Over how wide a spectrum of the animal kingdom can one make convincing homologies
based on expression domains of developmental genes? There are, at present, several attempts to use developmental genetic
data to establish homologies between body regions of animals with highly divergent body plans. Most notably, there has been
a revival of the once scorned proposal that some ancestral arthropod or annelid underwent an inversion of the dorsoventral
body axis and gave rise to the vertebrates. It will be interesting to see whether this relatively radical approach can be

substantiated by the rapidly advancing knowledge of developmental genes in a diversity of animal phyla.

The celebrated 1830 debate between Cuvier and Geoffroy
Saint-Hilaire, although it covered a variety of issues, was
triggered by their deep differences over how broadly the no-
tion of unity of composition (also termed unity of plan)
should be applied. In other words, the question was: Over
how wide a spectrum of the animal kingdom should one try
to establish homologies? Cuvier, for all his insistence on the
primacy of function over form, still recognized homologies
based on form — but only between animals with quite similar
overall body plans (e.g., between mammals and birds). In
contrast, Geoffroy had absolutely no reservations about com-
paring structures between animals with radically different
body plans (e.g., between vertebrates and molluscs). In the
opinions of Appel (1987) and Corsi (1988), the debate did
not resolve this issue clearly.

For a century and a half following the debate, biologists
continued to be divided into those taking comfort in the nar-
rower vision of Cuvier and those daring to consider the
broader, more exciting, but seemingly less reliable vistas of
Geoffroy. Most recently, striking advances in developmental
genetics have revived interest in this old controversy and
have suggested to some (like Gould, 1991 and Slack et al.,
1993) that Geoffroy was, after all, closer to the truth than
Cuvier.

In recent years, the evolutionary implications of data from
developmental genetics have been discussed under the rubric
of “developmental evolution” (“devo-evo” for short) or, more
soberly, “comparative molecular genetics” (Lawrence and
Morata, 1994). Such data can enrich evolutionary discussions
in several ways. For instance, one can construct family trees
of developmental genes (Schugart et al., 1989; Schubert et
al., 1993) or one can study developmental gene duplication
events in evolutionary lineages (Garcia-Fernandez and Hol-
land, 1994). The present paper, however, is concerned with

New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

only one way of utilizing developmental genetic data:
namely, the use of expression domains of developmental
genes as phenotypic characters for helping to establish ho-
mologies between body parts of distantly related animals.
This approach, although often employed during the last few
years, has not yet been thoroughly discussed in relation to
homology — not even in the most recent book on the latter
subject (Hall, 1994). Thus my present purposes are: (1) to
review the barest essentials of developmental genetics and
homology and (2) to illustrate our method by discussing two
examples from our recent work, which is relevant to the
origin and early evolution of the vertebrates. Coverage of
developmental genetics is strictly limited here to facts es-
sential for what follows. Readers wishing a fuller introduc-
tion to the field are referred to the references cited in this
section, which have been selected for readability and breadth
of view.

The last decade has witnessed the unexpected discovery
that the molecular machinery of development is remarkably
similar from one animal phylum to the next (De Robertis et
al., 1990; Holland, 1990, 1992; Lawrence, 1992; McGinnis
and Krumlauf, 1992; Jacobs, 1994). Several kinds of devel-
opmental genes and their protein products include structural
motifs that are highly conserved across a wide range of ani-
mal phyla—a concrete example would be the correspon-
dence of a particular homeotic gene in a fruitfly with a par-
ticular Hox gene of a vertebrate. Conservation is evident not
only at the level of base and amino acid sequences, but also
in the spatiotemporal patterns of gene expression in the em-
bryo. All this conservation of sequence and expression has
frequently been assumed to reflect a conservation of function.
This assumption very recently has begun to find experimental
support (McGinnis and Kuziora, 1994).

The best known example of a conserved motif in devel-

opmental genes 1s the homeobox (McGinnis, 1994), a se-
quence of 180 base pairs encoding a 60-amino acid homeo-
domain which is part of a protein that can bind to DNA and
act as a transcription regulator. The most conspicuous (al-
though not the only) function of homeobox genes is to spec-
ify positional identities to body regions along the anterior-
posterior axis of developing embryos. Two such homeobox
genes will be used as examples later in the present paper,
after a review of homology.

The most recent and synoptic review of homology (Hall,
1994) reveals that today’s biologists are more divided than
ever on most aspects of the subject, including its uses, con-
cepts, definitions, and recognition criteria. Therefore, a brief
review of homology will be useful here for clarifying what
my co-workers and I are doing and, just as importantly, what
we are not doing. I will sidestep some aspects of homology
such as its utility at the level of populations, communities
and zoogeographic zones. These topics are thoroughly ex-
plored in the chapters in Hall (1994).

Before discussing concepts and criteria of homology, it is
only fair to recognize a couple of general difficulties with
the subject. The most fundamental is the lack of consensus
about how the human mind functions in identity recognition.
Biologists are usually content to leave this issue to philoso-
phers and psychologists while getting on with the business
of comparing features of different organisms.

The second pervasive problem with homology concerns
biological identity. Unlike deductively proven mathematical
identities, inferences about biological identities do not have
probabilities of one. For example, when we homologize two
structures, the probability that we are correct ranges from
close to one (say for a mammalian femur versus a bird femur)
to considerably less than one (say for annelid setae versus
brachiopod setae). The lower the probability, the less com-
fortable we feel, and there is as yet no agreement on how
to quantify our discomfort. Simply passing the buck to a
computer rarely makes this problem disappear. Thus, like
much else in the historical sciences, homologies and the phy-
logenies built upon them may not be falsifiable in the strict
sense of Popper. Again, most biologists, in order to transact
any business at all, tend to leave this problem to the phi-
losophers.

The concepts and criteria of homology are easiest to review
in approximate chronological order. Up to the late 1700s,
the homology concept was intuitive (which is to say there
was no concept at all), Even so, some good homologies were
made: a well-known example was Belon’s comparison of
bird and human skeletons in 1555. No explanations were
offered, and no recognition criteria were stated.

From the late 1700s through the mid 1800s, the homology
concept was idealistic. The explanation was that structures
in different organisms are homologous because they corre-
spond to the same archetype = to the same Law of Nature
= to the same Idea in the mind of God. During this era, the
structures in question were usually those of adult, extant or-
ganisms, although features of embryos and extinct species
were sometimes included in the analysis. In comparing a
given structure in two different organisms, the chief criteria

NICHOLAS D. HOLLAND

for recognizing homology were usually considered to be the
following: (1) positional equivalence within the overall body
plan, e.g., in different groups of higher vertebrates, the tem-
poral bones have similar positions within comparable sys-
tems of other skull bones; (2) special quality, e.g., the testes
of all vertebrates agree in special features (like seminiferous
cysts and tubules) without necessarily having equivalent po-
sitions; (3) transition, such that, in a comparison between
two species, apparently dissimilar structures are recognized
having positional equivalence or special quality if united in
a series by transitional structures in additional species. It
should be understood that transition in this context does not
refer to the substitution of one base for another in a nucleic
acid. It should also be understood that each criterion 1s known
by a variety of names; however, this changes nothing, beyond
making some authors a little harder to read.

Since the mid 1800s, the dominant homology concept has
been historical. The basic explanation 1s that structures are
homologous because they correspond to an equivalent struc-
ture in a common ancestor. The chief recognition criteria
are the same three used during the era of idealistic homology.
To these three criteria, a fourth 1s sometimes added: this is
the congruency criterion, which means that each suspected
homology should be examined in the context of other lines
of relevant evidence (clearly, if one line of evidence contra-
dicts another, something has to be wrong).

Over the years, several amendments have been made to
the historical homology concept. As a result, the categories
of compared features have increased well beyond structural
parts of adult organisms. First came the addition of embry-
onic primordia, with the connotation that, in different ani-
mals, structures derived from an equivalent embryological
primordium are especially likely to be homologous. Later
additions included physiological and behavioral characters.
Most recently, there has been wide acceptance of molecular
traits as indicators of homology — originally at the level of
metabolic pathways and later at the level of base and amino
acid sequences in genes and their protein products, respec-
tively. At this point, it became important to distinguish be-
tween orthology (correspondence between molecules in two
different organisms) and paralogy (correspondence between
molecules within the same organism). Increasing the number
of categories of evidence for homology has raised the unre-
solved problem of what to do when different categories do
not support one another. For example, molecular biology sug-
gests lampreys and hagfishes should be grouped together,
but morphology indicates that lampreys are more closely re-
lated to gnathostomes than either is to hagfishes (Forey and
Janvier, 1994).

An additional development within the historical concept
of homology has been the augmentation of the criterion of
transition. First there has been the use of stratigraphic data
to give a time dimension and directionality for changes in
homologous structures, and second there has been an in-
creased insistence that transitional character states must be
functionally credible.

In recent years, even as the historical concept was being
refined, another concept, that of biological homology, was

HOMOLOGY, HOMEOBOX GENES AND EARLY EVOLUTION

introduced. This concept is built upon the notion of the epige-
netic landscape that Waddington (1940) originally proposed
in a purely developmental context. Waddington’s ideas, as
transplanted to the realms of homology and evolution, yield
the following explanation: developmental constraints con-
serve homologous traits by channeling development to per-
missible end-points separated by virtually impossible end-
points. Thus homologies should be underlain by (and, in the
thinking of some, should be equivalent to) networks of in-
ductive interactions that become very hard to alter even in
part. It is commonly, but by no means universally, proposed
that the chief criterion for recognizing biological homology
is the discovery of conserved networks of epigenetic inter-
actions during development (for a critical review of this no-
tion, see Hall, 1995). In all, the biological homology concept
has defied crisp definition so far, primarily because the un-
derlying notion of constraint is still so contentious.

At this point, it is useful to interpolate a paragraph to record
the current attitudes of cladists toward homology problems.
A few cladists think that they can somehow build a
cladogram without prior consideration of homologies. The
remaining cladists — for whom a data matrix of good ho-
mologies is the prerequisite for good tree building — hold
various combinations of beliefs, and I will mention only a
few recurring themes. In dealing with homology concepts,
some cladists prefer to avoid reference to ancestors (1) either
by adhering to the historical concept but reformulating it in
terms of synapomorphy (2) or by embracing the biological
concept, which they sometimes redefine in terms of conti-
nuity of information instead of gene networks. In dealing
with homology criteria, some cladists reject transition, evi-
dently because this criterion is partly grounded on paleon-
tological information that can sometimes indicate a series of
transitions along an unbranched segment of a lineage (1.e.,
anagenesis); such data may be considered inappropriate for
a strictly cladistic analysis based only on branching.

To return to the main stream of the present paper, I agree
with Minelli (1993) that more than one concept of homology
is probably defensible; indeed, the most appropriate concept
to use could be determined by the level of biological organi-
zation under consideration (Hall, 1994). | was brought up in
the tradition of the historical concept, but think the biological
concept is strengthened by recent advances in developmental
genetics and is probably superior for accommodating phe-
nomena of serial homology. Even so, for what follows, |
will use the historical and not the biological concept. My
reasons are strictly practical and not doctrinal. The problem
with the biological homology concept is that little if anything
has yet been discovered about regulatory gene networks in
animals of the greatest phylogenetic interest (e.g., monopla-
cophoran molluscs, peripatus, pterobranchs, appendicularian
tunicates, amphioxus, hagfishes, etc.). Thus, for my purposes,
I will fall back on the historical concept.

Here, then, is the crux of our approach. Within the frame-
work of the historical homology concept, we are using the
criterion of special quality sequentially at two different levels
of organization. The first level is molecular: at this point,
homologies between genes are identified from the overall

agreement of aligned sequences of bases or of the amino
acids encoded by them. The second level at which we use
the criterion of special quality is phenotypic: at this point,
spatiotemporal expression patterns of homologous genes, as
detected by the presence of their mRNA or protein products,
are used as phenotypic characters to indicate homologous
structures between species. It should be understood that such
gene expression data help establish homologies, not in iso-
lation, but in conjunction with other relevant characters.

Implicit in our approach is the assumption that there is a
one-to-one mapping among the genetic, epigenetic (gene net-
work), and phenotypic levels of organization for the particu-
lar genes we are studying. For instance, in the words of
Akam (1995), “Hox genes are not just markers for homology,
they are part of the mechanism that defines it.” This seems
to be so, but little can yet be said about such a mechanism
except that it probably involves at least the following two
features: first, Hox genes and their contiguous gene networks
appear to remain relatively stable over evolutionary time,
and, second, the modest changes that have occurred in Hox
expression domains and in Hox interactions with other genes
can alter phenotypes dramatically (Day, 1995). Obviously,
the assumption of one-to-one mapping may not hold for de-
velopmental genes when multiple hierarchical processes in-
teract strongly to produce a phenotype (Hinchliffe, 1994),
and in some other instances (Hall 1995).

My first example compares a jawless vertebrate (a lam-
prey) with gnathostome vertebrates; details of this work are
in Holland et al. (1993). The parts compared are specific
jaw muscles, and the genetic marker is the engrailed gene
(en). The most comprehensive comparison of en genes with
one another and with other homeobox genes is in Birglin
(1994). Only one en gene is known in lampreys (Holland
and Williams, 1990), while two or three paralogs of en have
been found in each major gnathostome group.

The first step in our homology recognition is a comparison
of amino acid sequences in the lamprey and mouse ev pro-
teins. Most of the amino acid sequence for the homeodomain
of lamprey en was determined by Holland and Williams
(1990). When this sequence is compared to the amino acid
sequences in engrailed homeodomains of other vertebrates
(sequence data from Scott et al., 1989), the amino acid iden-
tities range from 75%-80%; in contrast, only 45%-—S0%%
amino acid identities are found when homeodomains are
compared between lamprey en and any of the vertebrate Hox
genes. Thus, homology between the engrailed genes of lam-
prey and gnathostomes is established by the high proportion
of shared, identical amino acids.

The second step in our homology recognition begins with
the detection of the expression domain for lamprey ev. For
a series of lamprey embryos and larvae, en expression 1s
detected by exposing them to an antibody raised against
mouse en protein. Cells expressing the gene can be demon-
strated by immunohistochemistry, which produces a dark re-
action product in whole mounts and histological sections.
The first detectable em expression is seen at the stage of
early head development; at this time, the gene product ap-
pears in neural tube cells at the midbrain/ hindbrain boundary

FIGURE |. Early expression domains (stippled) of engrailed in
developing lamprey (above) and zebrafish (below). Abbreviations
are: FB forebrain; MB midbrain; HB hindbrain; Y yolk; WT future
velothyroideus muscle; LAP levator arcus palatini; DO dilator oper-
culi. Based on data from Hatta et al. (1990) and Holland et al. (1993).

and in some paraxial mesoderm cells at either side of the
mandibular arch. During the following week, the mesoderm
cells continue to express en while differentiating into the
velothyroideus muscles, which provide much of the power
stroke for the velum.

Figure | compares our data from lamprey embryos with
the expression domains of en that Hatta et al. (1990) dem-
onstrated in the head of zebrafish embryos. In both embryos,
there is a zone of en-expressing cells in the neural tube at
the midbrain/hindbrain boundary; this is not a very exciting
discovery, since both gross and fine neuroanatomy long ago
revealed that this region of the neural tube is homologous
between lampreys and gnathostomes. Far more intriguing is
the similarity between part of the myogenic mesoderm of
the embryonic mandibular arch of lampreys and zebrafish
(Fig. 1). Zebrafish en is expressed in mesenchyme cells of
an initially single premyogenic condensation, and expression
continues as they differentiate into two of the jaw muscles
(the levator arcus palatini and the dilator opercul). We pro-
pose that these en expression domains (in conjunction with
patterns of peripheral innervation) can be used as phenotypic
characters to help establish homologies between (1) the
velothyroideus of lampreys and (2) the levator arcus palatini
and dilator operculi of teleosts.

Our proposed homology between velar muscles of lam-
preys and some of the jaw muscles of gnathostomes is rele-
vant for the long-standing question of the phylogenetic origin
of the jaws of vertebrates. The traditional, textbook scenario,
formulated over a century ago by Balfour, derives the jaws
from elements of a relatively undifferentiated cranial arch in
the proximate common ancestor of the vertebrates. Recently,

NICHOLAS D. HOLLAND

the broad outline of this scenario has been retained by Mallatt
(1996), who has fleshed it out with a wealth of new detail.
He has interpreted our data on lamprey en to be consistent
with his scenario, which starts with a common ancestor that
lacks a muscularized velum: in the line leading to the living,
jawless vertebrates, there was a centripetal migration of the
most anterior circlet of superficial branchial constrictor mus-
cles to form the velum of jawless vertebrates; on the other
hand, in the line leading to the jawed vertebrates, these same
muscles stayed in place and took on additional functions in
suction feeding while retaining the primitive ventilatory func-
tion. Alternatively, our lamprey en data might also be used
to support a converse scenario for the origin of the vertebrate
jaws (Jollie, 1977; Forey and Janvier, 1993, 1994). For these
authors, the proximate, common ancestor of the vertebrates
had a muscularized velum that was retained during the evo-
lution of the jawless vertebrates, but gave rise to the jaws
of gnathostomes. It will be interesting to see whether any
further evidence will be obtained to support this challenge
to the traditional scenario for the origin of vertebrate jaws.

My second example compares two different subphyla of
the phylum Chordata: namely a cephalochordate, amphioxus
(Branchiostoma floridae), versus the vertebrates. The parts
compared are regions of the dorsal nerve cord, and the ge-
netic markers are Hox genes — specifically, AmphiHox3 in
amphioxus versus Hox genes of vertebrates. Details of clon-
ing, sequencing and in situ hybridization of the amphioxus
gene are in Holland et al. (1992a).

As the first step in our homology recognition, the amino
acid sequences of the homeodomains are compared between
AmphiHox3 of amphioxus and Hox genes of vertebrates (se-
quence data from Lonai et al., 1987; Scott et al., 1989; Hol-
land et al., 1992a). This comparison of homeodomain amino
acids between amphioxus and vertebrates gives the following
percentages of identities when AmphiHox3 is compared to
the following Hox genes: 92% for vertebrate HoxB3, 87%
for vertebrate HoxAd3, 87% for vertebrate HoxD3: and 60-
80% for all the Hox genes of vertebrates. In this example,
homology is established between AmphiHox3 and vertebrate
HoxB3 by the high percentage of identical amino acids in
the respective homeodomains. This homology 1s further sup-
ported by the corresponding position of the introns in both
genes and by the exceptionally long amino acid sequence
downstream from the homeodomain in both gene products.

The second step in our homology recognition begins with
the detection of the expression domain for amphioxus Am-
phiHox3. For a series of amphioxus embryos and larvae,
expression of this gene is found by in situ hybridization with
a riboprobe recognizing the MRNA message for AmphiHox3
in whole mounts of embryos and larvae. The first detectable
expression of AmphiHox3 is seen in posterior, undifferenti-
ated mesoderm and in the dorsal nerve cord of neurula em-
bryos a few hours after hatching. The anterior limit of neural
expression is at the boundary between the fourth and fifth
somites, where it remains for approximately the next week
of development. The somites are convenient fiducial marks
for judging the anterior boundary of expression of Amphi-
Hox3.

HOMOLOGY, HOMEOBOX GENES AND EARLY EVOLUTION

Sass

FiGuRE 2. Expression domains (diagonally hatched) of AmphiHox3 in an amphioxus embryo (above) and of HoxB3 in a generalized
vertebrate embryo (below). The mesodermal somites of the amphioxus embryo are numbered, as are the hindbrain rhombomeres of the vertebrate

embryo. Based on data from Holland et al. (1992a).

Figure 2 compares the expression domain of AmphiHox3
with the expression domain of its homologue (the equivalent
of mouse HoxB3) in a generalized vertebrate embryo. In the
vertebrate, the anterior limit of HoxB3 expression is halfway
between the rostral and caudal limits of the hindbrain (at the
boundary between rhombomeres four and five). We thus pro-
pose that the anterior limit of AmphiHox3 expression in am-
phioxus marks a location homologous to the middle of the
vertebrate hindbrain.

Our work on AmphiHox3 immediately raises the question
of the extent of the amphioxus hindbrain. Our recent, un-
published work on expression of the amphioxus engrailed
gene (AmphiEn) places the anterior limit of the hindbrain
far anteriorly in the central nervous system (a conclusion
already tentatively reached from a neurochemical study by
Holland and Holland, 1993). It thus seems likely that am-
phioxus embryos have a very large hindbrain; indeed, one
of my colleagues refers to them as “swimming hindbrains.”
The small percentage of the central nervous system rostral
to the proposed anterior limit of the hindbrain might be
largely or entirely a homologue of the vertebrate diencepha-
lon (according to neuroanatomical data in Lacalli et al.,
1994).

If one accepts that the expression domains of developmen-
tal genes can help establish homologies, it should be possible
to correlate many parts of the central nervous system between
amphioxus and vertebrates. This can be done by demonstrat-
ing expression domains of other developmental genes (ad-
ditional Hox genes, Krox, Pax, Emx, Otx, Dix, and the like)

already known to mark specific regions of the vertebrate
brain (Holland et al., 1992b). For amphioxus, this approach
could reveal the posterior limit of the hindbrain and help
establish the identity of the small region of the central nerv-
ous system lying anterior to the hindbrain.

The presence of a large hindbrain in amphioxus runs
counter to the usual view that the cephalochordate central
nervous system comprises an extremely long spinal cord and
a minute, anterior brain (the cerebral vesicle). Even from the
few results we have in hand, it seems likely that the verte-
brate brain did not evolve solely from the paltry cerebral
vesicle of an amphioxus-like ancestor. Instead, vertebrates
may have evolved from a proximate ancestor with a surpris-
ingly large brain.

The animals compared in each of the examples above are
much more distantly related than Cuvier would have liked,
although they are all members of the phylum Chordata. In
principle, however, our approach could be extended to com-
pare body regions between even more distantly related ani-
mals — if their overall body plans are relatively similar. For
a concrete example, the discovery of Brachyury expression
in the stomochord of the phylum Hemichordata would
strongly indicate that this anterior gut diverticulum is the
homolog of the notochord of the phylum Chordata. Akam
(1995) has criticized the use of single developmental genes
as indicators of homology; however, there is little doubt that
demonstration of Brachyury in the stomochord would be the
decisive piece of information for settling the century-old con-
troversy over whether hemichordates have notochords.

We now come to the unresolved question of what happens
as we compare animals with increasingly diverse body plans.
Slack et al. (1993) were the first to push the envelope. They
compared the later (phylotypic stage) embryos of widely di-
vergent phyla of metazoans and extracted a lowest common
denominator “zootype.” This is a simple, worm-like form
with its anterior-posterior axis divided into regions distin-
guished by the expression of different Hox genes during de-
velopment. Slack et al. (1993) suggested that the order of
the expression domains of these genes is a very ancient ar-
rangement, probably already in place in the common ancestor
of all bilateral animals. This conclusion appears valid, prob-
ably because the authors avoid precise anatomical identifi-
cations of the body parts being compared.

Another comparison of animals with very diverse body
plans concerns the eye. Unlike a vague body region of a
zootype, the eye is a very well-defined organ. Pax-6 genes
(which include not only a homeobox, but also another con-
served motif, the paired box) have turned out to be the master
control genes specifying eyes in a wide variety of inverte-
brates and vertebrates (Quiring et al., 1994; Halder et al.,
1995). Pax-6 itself is not a major repository for the devel-
opmental blueprint of eyes (although Gould 1994 seems to
imply that it is): instead, this gene is the upstream trigger
for developmental gene cascades that result in eye develop-
ment. Thus, as clearly expressed by Patel (in Barinaga, 1995),
the Pax-6 results could well mean that all animal photore-
ceptors are homologous as cells, but not that all animal eyes
are homologous as organs. It is too early to be certain, but,
when the overall gene cascades downstream from Pax-6 are
elucidated for eye development, there could be substantial
differences from one phylum to the next.

To date, the most interesting and important attempt to ho-
mologize body regions in widely diverse animal phyla is
focused on the possible reversal of the dorsal and ventral
axes during the origin of the chordates. Geoffroy Saint-
Hilaire was the first to propose such a notion —a decade
before his debate with Cuvier (details in Cahn, 1962). This
idea, which has been in great disrepute for most of the twen-
tieth century, has now been revived by several developmental
geneticists (Gehring in Wright, 1994; Arendt and Niubler-
Jung, 1994; Nibler-Jung and Arendt, 1994; Holley et al.,
1995; Travis, 1995). Although data on the expression do-
mains of a few isolated genes would not be good grounds
for stampeding us all the way to Geoffroy’s extreme position,
there is more going on here: interactions within networks of
dorsal-ventral patterning genes are beginning to be discov-
ered. It seems likely that the body region homologies being
proposed will become more convincing in proportion to the
number of interacting genes elucidated. Just how large a con-
stellation of gene interactions would have to be elucidated
in order to demonstrate a convincing homology remains an
unanswered question. This 1s a question that bids fair to pre-
occupy biologists for years to come.

NICHOLAS D. HOLLAND

The recent developmental genetic comparisons of animals
with highly divergent body plans forcefully raise the issue
of what to do if structures traditionally considered to result
from convergent evolution should be shown to develop under
the control of homologous genes and gene networks. As if
this problem weren’t knotty enough, an even more radical
step has recently been taken by Minelli and Schram (1994).
They have proposed that: (1) Especially important combina-
tions of organ-specifying developmental genes occur at a
relatively few, fixed locations (“hot spots”) along body axes.
(2) The hot spots are the same within major groupings of
animals (roughly encompassing a few phyla each). (3) A
given hot spot might become coupled to different cascades
of down-stream genes in different animals (Krumlauf [1994]
admits the possibility of such differential coupling, but
stresses that it lacks experimental support at present). (4)
Phenotypically dissimilar structures (say an appendage ver-
sus a gonoduct) resulting from a similar hot spot should be
considered “positional homologies.” As a result, structures
never before mentioned in the same breath — not even as
convergences — could be related as positional homologies.
These ideas are not supported by the recent work of Burke
et al. (1995), who found instances where Hox expression
marks phenotypic characters and not fixed axial positions.
However, much more needs to be discovered about devel-
opmental genetics before the validity of positional homology
can be adequately tested.

In conclusion, even the relatively scanty developmental
genetic data now in hand can indicate body part homologies
when the animals being compared have body plans that are
only moderately divergent. Within these limits, there is much
useful work to be done, as exemplified here by our results
from lampreys and amphioxus.

However, there are more radical ways to proceed in de-
velopmental evolution. The risks are greater, but so are the
potential rewards. As more is learned about developmental
gene cascades in more kinds of animals, we may be able to
compare organisms with markedly different body plans and
make convincing homologies between organs or body re-
gions. Clearly, homologies of just this sort would be the
most useful ones for working out the broad outlines of the
evolutionary relations among the animal phyla.

It is an intriguing possibility that a more perfect knowledge
of comparative molecular genetics —not just for a few
“model system” animals, but for a healthy variety of meta-
zoan phyla — will give us a much clearer picture of the true
topography of the tree of animal life, so long sought. The
new genetic data have started being used to push the principle
of unity of composition to the sort of extremes that once
delighted Geoffroy Saint-Hilaire. The ultimate success of this
radical approach cannot yet be foreseen clearly (it is even
possible that Cuvier might someday have the last laugh).
Even so, the mood of developmental evolutionists is currently
one of anticipation and excitement.

HOMOLOGY, HOMEOBOX GENES AND EARLY EVOLUTION
Acknowledgments

This manuscript was improved by the critical comments
of Linda Z. Holland, Jon Mallatt, Margaret McFall-Ngai, an
anonymous (and very astute) reviewer, and my fellow par-
ticipants in the Symposium on New Perspectives on the His-
tory of Life at the California Academy of Sciences. This
work was supported in part by NSF Research Grant IBN
92-21622.

Literature Cited

Akam, M. 1995. Hox genes and the evolution of diverse body plans.
Phil. Trans. Roy. Soc. Lond. B 349:313-319.

Appel, T. 1987. The Cuvier-Geoffroy Debate. Oxford Univ. Press,
Oxford.

Arendt, D. and K. Niibler-Jung 1994. Inversion of dorsoventral axis?
Nature 371:26.

Barinaga, M. 1995. Focusing on the eveless gene. Science
267:1766—1767.

Biirglin, T. R. 1994. A comprehensive classification of homeobox
genes. Pp. 27-71 in Guidebook to the Homeobox Genes (D.
Duboule, ed.). Oxford Univ. Press, Oxford.

Burke, A. C., C. E. Nelson, B. A. Morgan and C. Tobin. 1995. Hox
genes and the evolution of vertebrate axial morphology. Devel-
opment 121:333-346.

Cahn, T. 1962. La Vie et ’Oeuvre d’Etienne Geoffroy Saint-Hilaire.
Presses Universitaires de France, Paris.

Corsi, P. 1988. The Age of Lamarck: Evolutionary Theories in
France 1790-1830. Univ. California Press, Berkeley.

Day, S. 1995. Invasion of the shapechangers. New Scientist 148
(No. 2001):30-35.

De Robertis, E., G. Oliver, and C. V. E. Wright. 1990. Homeobox
genes and the vertebrate body plan. Sci. Amer. 263 (1):46—52.

Forey, P. and P. Janvier 1993. Agnathans and the origin of jawed
vertebrates. Nature 361:129—134.

———. 1994. Evolution of the early vertebrates. Amer. Sci.

82:554-S65.

Garcia-Fernandez, J. and P. W. H. Holland. 1994. Archetypal or-
ganization of the amphioxus Hox gene cluster. Nature 370:563—
566.

Gould, S. J. 1991. Of mice and mosquitoes. Nat. Hist. 100 (7):12—
20.

1994. Common pathways of illumination. Nat. Hist.
103(2):10—20.

Halder, G., P. Callaerts and W. J. Gehring. 1995. Induction of ec-
topic eyes by targeted expression of the eyeless gene in Droso-
phila, Science 267:1788—1792.

Hall, B. K. 1994. Introduction. Pp. 1-19 in Homology: The Hier-
archical Basis of Comparative Biology (B. K. Hall, ed.). Aca-
demic Press, San Diego.

. 1995. Homology and embryonic development. Evol. Biol.
28:1—37.

Hatta, K., T. F. Schilling, R. A. BreMiller and C. B. Kimmel. 1990.
Specification of jaw muscle identity in zebrafish: correlation with
engrailed-homeoprotein expression. Science 250:802—805.

Hinchliffe, J. R. 1994. Evolutionary developmental biology of the
tetrapod limb. Development 1994 Supplement:163—168.

Holland, N. D. and L. Z. Holland. 1993. Serotonin-containing cells
in the nervous system and other tissues during ontogeny of a
lancelet, Branchiostoma floridae. Acta Zool. (Stockh.) 74:195—
204.

Holland, N. D., L. Z. Holland, Y. Honma and T. Fujii. 1993. En-
grailed expression during development of a lamprey, Lampetra
japonica: a possible clue to homologies between agnathan and
gnathostome muscles of the mandibular arch. Develop. Growth
and Differ. 35:153—160.

Holland, P. W. H. 1990. Homeobox genes and segmentation: co-
option, co-evolution, and convergence. Seminars Dev. Biol.
1:135—145.

evolution.

1992. Homeobox vertebrate

BioEssays 14:267-273.

genes in

Holland, P. W.H., L. Z. Holland, N. A. Williams and N. D. Hol-
land. 1992a. An amphioxus homeobox gene: sequence conser-
vation, spatial expression during development and insights into
vertebrate evolution. Development 116:653—661.

Holland, P. W. H., P. Ingham and S. Krauss. 1992b. Mice and flies
head to head. Nature 358:627—628.

Holland, P. W. H. and N. A. Williams. 1990. Conservation of en-
grailed-like homeobox sequences during vertebrate evolution.
FEBS Letters 277:250—252.

Holley, S. A., P. D. Jackson, Y. Sasal, B. Lu, E. M. De Robertis,
F. M. Hoffmann and E. L. Ferguson. 1995, A conserved system
for dorsal-ventral patterning in insects and vertebrates involving
sog and chordin. Nature 376:249-253.

Jacobs, D. K. 1994. Developmental genes and the origin and evo-
lution of Metazoa. Pp. 537-549 in Molecular Ecology and Evo-
lution: Approaches and Applications (B. Schierwater, B. Sterit,
G. P. Wagner, and R. DeSalle, eds.). Birkhauser Verlag, Basel.

Jollie, M. 1977. Segmentation of the vertebrate head. Amer. Zool.
17:323-333.

Krumlauf, R.
78:191-201.

1994. Hox genes in vertebrate development. Cell

Lacalli, T.C., N. D. Holland and J. E. West. 1994. Landmarks in
the anterior central nervous system of amphioxus larvae. Phil.
Trans. Roy. Soc. Lond. B 344:165—185.

70
Lawrence, P. A. 1992. The Making of a Fly. Blackwell, London.

Lawrence, P. A. and G. Morata. 1994. Homeobox genes: their func-
tion in Drosophila segmentation and pattern formation. Cell
78:181—189.

Lonai, P., E. Arman, H. Czosnek, F. H. Ruddle, and C. Blatt. 1987.
New murine homeoboxes: structure, chromosomal assignment,
and differential expression in adult erythropoiesis. DNA 6:409—
418.

Mallatt, J. 1996. Ventilation and the origin of jawed vertebrates: a
new mouth. Zool. J. Linnean Soc. In press.

McGinnis, W. 1994. A century of homeosis, a decade of ho-
meoboxes. Genetics 137L 607-611.

McGinnis W. and R. Krumlauf. 1992. Homeobox genes and axial
patterning. Cell 68:283—302.

McGinnis, W. and M. Kuziora. 1994. The molecular architects of
body design. Sci. Amer. 270 (2):58—66.

Minelli, A. 1993. Biological Systematics: the State of the Art.
Chapman and Hall, London.

Minelli, A. and F. R. Schram. 1994. Owen revisited: a reappraisal
of morphology in evolutionary biology. Buijdragen tot de
Dierkunde 64:65—74.

Nibler-Jung, K. and D. Arendt. 1994. Is ventral in insects dorsal
in vertebrates? A history of embryological arguments favouring

NICHOLAS D. HOLLAND

axis inversion in chordate ancestors. Roux’s Arch. Dev. Biol.
203:357-366.

Quiring, R., U. Waldorf, U. Kolter and W. Gehring. 1994. Homol-
ogy of the eyeless gene of Drosophila to the small eye gene in
mice and aniridia in humans. Science 265:785—789.

Schubert, F.R., K. Nieselt-Struwe, and P. Gruss. 1993. Anten-
napedia-type homeobox genes have evolved from three precursors
separated early in metazoan evolution. Proc. Natl. Acad. Sci. USA
90:143—147.

Schugart, K., C. Kappen, and F. H. Ruddle. 1989. Duplication of
large genomic regions during evolution of vertebrate homeobox
genes. Proc. Natl. Acad. Sci. USA 86:7076—7071.

Scott, M. P., J. W. Tamkum, and G. W. Harzell. 1989. The structure
and function of the homeodomain. Biochim. Biophys. Acta
989:25-48.

Slack, J. M. W., P. W.H. Holland, and C.F. Graham. 1993. The
zootype and the phylotypic stage. Nature 361:490-492.

Travis, J. 1995. The ghost of Geoffroy Saint-Hilaire. Frog and fly
genes revive the ridiculed idea that vertebrates resemble upside-
down insects. Science News 148:216—218.

Waddington, C. H. 1940. Organizers and Genes. Cambridge Univ.
Press, Cambridge.

Wright, D. 1994. Evolution of the eye: how low does it go? J.
NIH Research 6:36—-39.

BRAIN HETEROCHRONY AND ORIGIN OF THE MAMMALIAN MIDDLE EAR

Timothy Rowe

Department of Geological Sciences and
Vertebrate Paleontology Laboratory
University of Texas at Austin
Austin, Texas 78712

The mammalian middle ear forms a distinctive chain of tiny ossicles whose parallel histories in ontogeny and phylogeny are
among the most famous in comparative biology. During pre-mammalian history the auditory chain was attached to the
mandible, where it functioned in sound transmission to the inner ear. In mammals ancestrally the chain was torn free from
the mandible and displaced to a new position behind the jaw. In early mammalian ontogeny the auditory chain begins
development as a part of the mandible that is later torn free and displaced backward, recapitulating the evolutionary trans-
formation. Participation by mandibular elements in auditory transmission predates the origin of mammals by more than 100
million years; what is distinctly mammalian is that the mandibular elements become detached from the jaw and repositioned
behind it. Two competing theories have attempted to account for this transformation. An evolutionary hypothesis argues that
natural selection for improved high frequency audition is the mechanism, while a developmental hypothesis contends that
ontogenetic onset of functionality in jaw muscles is the driving mechanism. Neither hypothesis accounts for both the evolu-
tionary and developmental transformations, or for the repositioning of the ossicles behind the jaw.

Phylogenetic analysis indicates that the distinctive inflated mammalian neocortex arose at the same time that the middle ear
became detached from the jaw, in the last common ancestor of extant mammals. A study of cranial development in didelphid
marsupials using high resolution X-ray CT, histological, and cleared and stained specimens implicates differential growth of
the brain in detachment and repositioning of the ossicles. In early ontogeny the brain is a hydrostat that mechanically loads
and displaces surrounding tissues, and in mammals it grows to unprecedented size. The ear ossicles approach their mature
size during the third week of postnatal development while still attached to the jaw and participating in a continuous arcade
of elements extending from the fenestra vestibuli to the mandibular symphysis. The brain continues to grow for nine additional
weeks and in the process it bursts the arcade. As the circumference of the growing brain expands, the ossicular chain is
torn away from the mandible and carried backwards to its adult position behind the jaw. Unlike the competing hypotheses,
the geometry of the growing brain accounts for detachment of the auditory chain from the mandible in both ontogeny and
phylogeny, for the precise path of subsequent posterior displacement of the auditory chain during development, and for the
timing and extent of this movement. A heterochronic increase in the rate and duration of brain development, which arose

in Mammalia ancestrally, may have been the driving force behind the origin of the distinctive middle ear.

Introduction

The study of evolutionary morphology is more than a cen-
tury old, yet one might argue that we remain largely ignorant
of the mechanisms of morphological change that have op-
erated historically. While many potential mechanisms have
been identified and some studied experimentally, very few
are yet mapped onto phylogenies and hence few historic in-
stances of transformation are fully explained. This situation
promises to improve with the emergence of the new disci-
pline of evolutionary developmental biology (Hall, 1992;
Hanken, 1993; Hanken and Hall, 1993; Wake et al., 1993,
1996), and as it becomes more fully integrated with phylo-
genetic systematics. The recognition of heterochrony requires
an explicit, corroborated phylogenetic framework and it is
this point that makes phylogenetic systematics fundamental
to understanding the evolution of development (Fink, 1982;
Kluge, 1988). Together, phylogenetic systematics and evo-
lutionary developmental biology afford means to recognize
episodes of heterochrony and heterochronic cascades, to dis-
criminate between genetic and epigenetic factors controlling
development, and to map onto cladograms these hierarchical
agents of change as they have operated historically.

An illustration of how these disciplines might be integrated
can be found in a problem involving the origin of the mam-
malian middle ear (Rowe, 1996), in what is among the most

New Perspectives on the History of Life
Editors. M. T. Ghiselin and G. Pinna

famous transformations in comparative anatomy. The middle
ear in extant mammals forms a chain of ossicles that hangs
suspended from beneath the adult cranium and comprises
one of the most distinctive osteological characters of mam-
mals. The parallel ontogeny and phylogeny of these bones
is one of the most celebrated recapitulations known (Goo-
drich, 1930; de Beer, 1958). The middle ear bones began
their phylogenetic histories as hearing ossicles while located
in an ancient position extending between the fenestra
vestibuli, their point of connection to the inner ear, and the
dentary bone of the lower jaw. The ear ossicles thus partici-
pated in a continuous arcade of elements extending from the
mandibular symphysis to the cochlear housing of the skull.
The craniomandibular joint was formed between two bones
in the chain, the quadrate and articular, which served the
dual functions of hearing and feeding (Allin, 1975, 1986;
Bramble, 1978; Crompton and Parker, 1978; Kemp, 1982;
Kermack and Kermack, 1984). Over a 100 million year span
of pre-mammalian history the middle ear ossicles were grad-
ually reduced in size while the dentary was enlarged until
it came to participate in the cramiomandibular joint. In the
next step of this history, coinciding with the origin of the
“crown group” Mammalia, hearing and feeding were decou-
pled as the chain of ossicles became detached from the man-
dible. The dentary bone was the only element remaining in
the lower jaw, and the craniomandibular joint was established

solely between the dentary and squamosal bones. During this
transformation, the morphology of the ear ossicles and their
anatomical relationships to one another were largely con-
served, but as a group they migrated to a new location en-
tirely behind the condyle of the dentary. Detachment of the
ossicles from the mandible produced the condition that oc-
curs universally among adult mammals and that, under the
typological practices of Linnean taxonomy, was widely re-
garded as the definitive mammalian character (Olson, 1959;
Simpson, 1959). Despite its importance, the mechanism caus-
ing this evolutionary detachment of the auditory ossicles
from the jaw and their backward displacement has remained
poorly understood.

In the early ontogeny of extant mammals several of the
middle ear bones differentiate and begin to grow in their
primitive positions along the mandible and, for a time in
early development, there is a continuous chain of cartilages
extending from the oval window to the mandibular sym-
physis. Later, the ossicular chain separates from the mandibu-
lar arch and moves backwards from the jaw to assume its
derived position suspended solely by the cranium in a new
location entirely behind the mandible. Ontogeny thus reca-
pitulates phylogeny in what would seem to be a highly un-
likely transformation, the detachment of the ossicular chain
from the mandible and its repositioning in a new location
behind the jaw (Toeplitz, 1920; de Beer, 1937, 1958; Rowe,
1988: Filan, 1991).

The evolutionary transformation from a “mandibular ear”
(suspended between cranium and dentary) to a “cranial ear”
(suspended only from the cranium) involved significant re-
design of the most intricate regions of the skull. If the ear
functioned for 100 million years while attached to the man-
dible, why did it detach and shift to a new location? Why
is this transformation recapitulated in the ontogeny of extant
mammals? My goal in this study is to describe the morpho-
genesis of detachment and repositioning of the chain of mid-
dle ear ossicles in ontogeny and phylogeny. Although gen-
erally viewed as the culmination of a long, gradual
evolutionary history, | argue later that the episode of detach-
ment occurring in the last common ancestor of extant mam-
mals was qualitatively different from the preceding 100 mil-
lion year history of ossicular reduction.

Two hypotheses, one evolutionary and one developmental,
have attempted to account for the detachment of the ossicles.
The evolutionary hypothesis (Allin, 1975) views pre-mam-
malian history as being shaped by selection for high fre-
quency hearing. It views the detachment of the ossicular
chain from the mandible as merely an extension of this trend,
but it says nothing of the developmental mechanism that
might have engineered this transformation. The developmen-
tal hypothesis (Herring, 1993a; Maier, 1987), on the other
hand, argues that the onset of functionality of the jaw muscles
tears the chain away from the mandibular arch but it does
not attempt to describe this transformation in an evolutionary
framework. Neither hypothesis addresses both the develop-
mental and phylogenetic transformations, however, nor do
they explain the repositioning of the auditory chain to its
new location behind the craniomandibular joint. Are the

TIMOTHY ROWE

evolutionary and developmental hypotheses complementary,
or are they mutually exclusive? Does some other single
mechanism address both the ontogenetic and phylogenetic
transformations?

Answers to these questions may lie not so much in the
ear and cramiomandibular joint, where they are usually
sought, as in the developmental and phylogenetic history of
adjacent parts, particularly the brain. The histories of the
mammalian middle ear and brain were believed to be largely
independent of each other and to be the evolutionary products
of separate morphogenetic mechanisms, an image com-
pounded in the paleontological literature by assertions of con-
vergent evolution in both regions. But unrecognized asso-
ciations between the brain and middle ear emerge by
mapping the variable features of both regions onto a cor-
roborated phylogeny of mammals and their closest extinct
relatives (Figs. 1, 2). These associations manifest the hier-
archy of heterochrony and implicate a single cascading
mechanism in both the ontogenetic and phylogenetic trans-
formations of the mammalian middle ear.

Materials

This study of ontogeny and phylogeny was based upon
osteological preparations of adult and developmental speci-
mens of a diversity of mammalian species, and examination
of the major collections of synapsid fossils in North America,
Europe, Russia, and South Africa. The principal source of
developmental information was a densely sampled growth
series for the extant didelphid marsupial Monodelphis do-
mestica. Didelphids are among the least-encephalized of liv-
ing mammals and most closely resemble the ancestral mam-
malian condition in many features pertinent to the present
study (Jerison, 1973; Reig et al., 1987). A growth series of
more than 200 individuals was obtained from the Southwest
Research Foundation (San Antonio, Texas). In Mono-delphis
domestica, the gestation period is 14-15 days (Fadem et al.,
1982: Kraus and Fadem, 1987). The life span of Monodelphis
domestica is approximately 3 to 4 years. The term “adult”
in this case refers to “retired breeders” that were shipped
without precise age data by Southwest Research Foundation
but with the general description that retired breeders range
in age from 9 to 36 months. Specimens dating from postnatal
days 0, 1, 10, 15, 27, and 36 were serially sectioned using
conventional histological techniques and stained with azo-
carmine. Approximately 100 specimens dating from day 0
through adults were cleared and double stained for cartilage
and bone, and dried skeletons dating from postnatal day 27
through adults were also prepared.

Serially sectioned embryos of Dide/phis documenting the
earliest stages of skeletal condensation (stages 32-35 of
McCrady, 1938) were generously provided by the Wistar
Institute. These specimens comprise a small fraction of the
extensive collection described in McCrady’s (1938) classic
monograph on embryology of the opossum. Most of this once
preeminent collection was tragically lost in the 1950s, but a
few sets of serial sections survived and were sent to me by
Wistar. I located some additional fragments of the collection

BRAIN HETEROCHRONY

in the National Museum of Medicine. The stains are now
badly faded on nearly all surviving slides, but the prepara-
tions are still useful for studying the early phases of skele-
togenesis; catalog numbers indicate that they include several
of the sets of sections used by McCrady to define the de-
velopmental stages of Didelphis.

To augment conventional developmental preparations,
complete three-dimensional data sets of dried Monodelphis
skulls dating from day 27 through old age were generated
using an ultra high resolution industrial X-ray CT scanner
(Rowe et al., 1993; 1995). This tool can exceed the resolution
of medical CT scanners by two orders of magnitude and it
produced exceptional imagery of complete Monodelphis cra-
nia in 100 uw thick consecutive serial sections. A complete
3-D data set of imagery was generated for each of 5 skulls
along sagittal, coronal, and transverse axes. A comparative
framework for studying the CT imagery of Monodelphis was
provided by an earlier study (Rowe, et al., 1993, 1995) in
which a 3-D data set of CT imagery for the extinct synapsid
Thrinaxodon liorhinus was generated in 200 consecutive
serial sections along the three orthogonal axes. Thrinaxodon
(Figs. 1, 2) has long been of central interest in early mam-
malian history because it preserves much of the primitive
morphology that we might expect to have been present in a
distant ancestor of mammals. The opportunity to compare
serial sections of individual specimens simultaneously along
different axes while handling the intact specimens themselves
offered an exceptionally rich opportunity to visualize all de-
tails of complex three-dimensional morphology in comparing
the derived Monodelphis with its more primitive relative
Thrinaxodon. The Thrinaxodon specimen was generously
made available by the Museum of Paleontology, University
of California, Berkeley (sp. no. UCMP 40466).

Systematic Framework

The systematic framework of this analysis was critical to
its outcome. The following discussion is based on the un-
derstanding that mammals are the sister group of other extant
amniotes (Fig. 1), a conclusion that rests upon analysis of
developmental and adult morphology of both hard and soft
tissues in a series of phylogenetic tests that included both
extinct and extant taxa (Gauthier et al., 1988, 1989; Gauthier,
1994). The term Mammalia is a node-based name (de Quel-
roz and Gauthier, 1990, 1992, 1994) for a clade whose mem-
bership derives from ancestry rather than “defining” charac-
ters. The name is used in reference to the taxon stemming
from the last common ancestor of extant mammalian species
(Rowe, 1988, 1993: Rowe and Gauthier, 1992), what is
sometimes referred to as the “crown group” Mammalia.

Two additional taxa are referred to below that include
mammals and some of their extinct relatives (Figs. 1, 2).
The term Cynodontia is another node-based name referring
to the taxon stemming from the last common ancestor shared
by Mammalia and the extinct Late Permian taxon Procyno-
suchus (Kemp, 1979, 1980; Rowe, 1993). Cynodonts thus
include mammals and their closest extinct outgroups. The
extinct taxon Thrinaxodon is a basal member of Cyno-

—
ios)

dontia whose anatomy is almost uniformly plesiomorphic.
Thrinaxodon has been of interest to paleontologists for more
than a century in understanding the 100 million years of
history immediately prior to the origin of mammals (Rowe,
1993; Rowe, et. al., 1993, 1995). The term Synapsida refers
to a still more inclusive taxon (Fig. 1). Synapsida is a stem-
based name (de Queiroz and Gauthier, 1990, 1992, 1994)
for the taxon that includes mammals and all extinct taxa
closer to mammals than to other extant tetrapods (Gauthier
et al., 1988; Rowe, 1988).

The phylogeny of mammals and their extinct relatives has
received a great deal of attention over the last century and
it was one of the first segments of Vertebrata to be studied
phylogenetically (McKenna, 1975). During the last decade,
a number of independent analyses of early mammalian phy-
logeny were conducted (Gauthier et al., 1988; Rowe, 1988,
1993; Wible, 1991; Wible and Hopson, 1993) using taxon
morphological-character matrices designed for analysis with
maximum parsimony software such as PAUP (Swofford,
1986-1994), MacClade (Maddison and Maddison, 1992),
and HENNIG 86 (Farris, 1986). Although these studies
reached different conclusions on certain points of relationship
among extinct taxa, the results for all extant and most extinct
taxa were identical. The studies also disagreed in certain
judgments on character independence that caused different
authors to split or lump suites of cranial features along dif-
ferent lines. This disagreement 1s significant from systematic
and morphological standpoints, but the conflicting decisions
on how to score characters did not affect the topology of
the most parsimonious trees generated in the separate studies.
In fact, in two batteries of tests (Rowe, 1988, 1993), char-
acters of the skull could be entirely removed from the data
matrix and the postcranial data alone recovered the same
tree as did the complete skeletal data set for the taxa of
interest to the present study. The trees in Figures | and 2
show only points of relationship that are consistent with all
analyses, and they provide the systematic context in which
the histories of the middle ear and brain are discussed below.
Readers are referred to the original analyses for details of
phylogenetic methodology.

Paleontologists have long maintained that both the mam-
malian middle ear (Olson, 1959; Simpson, 1959; Kermack
and Kermack, 1984; Allin, 1986; Miao, 1991) and inflated
brain (Kielan-Jaworowska, 1986; Miao, 1991) evolved con-
vergently among synapsids. The genealogy supporting this
view was developed with phenetic methods which treated
Mammalia as an evolutionary grade and held that participa-
tion by the dentary and squamosal bones in the cranioman-
dibular joint constituted achievement of that grade (Rowe,
1993). Under the phenetic paradigm, the Late Triassic-Early
Jurassic fossils Morganucodon and Sinoconodon were
viewed as the oldest mammals because they are the oldest
fossils that have a load-bearing dentary-squamosal joint, and
their anatomy was taken to reflect the ancestral states of
mammalian characters. Because they retain a mandibular ear
and an uninflated brain, it followed that the ancestral
mammal did as well (Patterson and Olson, 1961; Edinger,
1964; Hopson, 1979; Crompton and Jenkins, 1973; Jerison,

(CH61 S86) MOY
pur (gg6]) ‘Je 3a JaryNeD Aq sasdyeur sHouaso[Ayd UO pose *s[PUILUPLU JO SOANLIAI JOUNXS payoaas pue spodesja) juerxa yo sdnosd sofeut ayy Jo AuasojAyd “| UNO

TIMOTHY ROWE

epodeja |
ueIUOASG
P]OIULUY SNOJA}IUOGUED

episdeuss

4 UOPOsOW!G UBILUed
Shy hac >,
eluopoUuAy
{UOPOXeUY oIssel |
yt
SOULIOJEIPWUWey

juopoonuebioy olsseune
D Bs — snosoelalg
Aveo,
BI/eUWUe YY eINdaYy BIGiYduly jUs08Y

ee pss SIYAEp!G ae ee

| mn
4) —

SMUNIION

BRAIN HETEROCHRONY

S
S »
» x 3
ey we < ~~
Oo? (2) & s~
< - \ SN
~\ o e @
& ee SC .
g S nn)

Cynodontia

~
nN

x
“SO 3
& & < ‘ S Na
x& \) x ~ x
few OF o SX &
ww) < <2 J

Mammalia
Mammaliaformes

Mammaliamorpha

Eucynodontia

FIGURE 2. Phylogeny of the major groups of living mammals and some of their closest extinct relatives among Cynodontia. The topology
among these taxa Is consistent with the results of separate analyses by Kemp (1983), Gauthier et al. (1988), Rowe (1988, 1993), Wible (1991),

and Wible and Hopson (1993).

1973, 1990). Consequently, the inflated brain and “cranial
ear must have evolved convergently in the lineages contain-
ing modern monotremes and therians after the two groups
diverged from their last common ancestor.

In contrast, the more recent phylogenetic outlook views
Mammalia as a clade, a position corroborated by the many
features from all systems that distinguish mammals from
other extant taxa (e.g., appendix B in Gauthier, et al., 1988).
Additionally, there are extensive lists of synapomorphies
from all parts of the skeleton based on analyses of extant
species and fossils (Gauthier et al., 1988, 1989: Rowe, 1988,
1993; Wible, 1991; Wible and Hopson, 1993; see also Kemp,
1983, Zeller, 1993). The analyses concur that monotremes
and therians are more closely related to each other than to
Morganucodon or Sinoconodon, and that the latter two taxa
are consecutive plesiomorphic outgroups to Mammalia (Figs.
12):

Phylogeny of the Middle Ear Ossicles

Early in synapsid history, the bones adjacent to the cra-

momandibular joint (CMJ) undertook the new function of

transmitting airborne sound vibrations to the inner ear while
maintaining their primitive structural role in the masticatory
system (Allin, 1975, 1986; Crompton and Parker, 1978; Gau-
thier et al., 1988, 1989: Kemp, 1982; Kermack and Kermack,

1984). An unbroken chain of bones extended from the fenes-
tra vestibuli to the symphysis of the mandibles and at first
“mandibular” hearing was probably restricted to low frequen-
cies owing to the massiveness of all bones in the transmission
pathway. Vibrations received by the mandible reached the
inner ear via the articular and quadrate, which formed the
CMJ, and from the quadrate via a massive stapes to the fenes-
tra vestibuli (Fig. 3). A rich fossil record documents the grad-
ual increase in relative size of the main tooth-bearing ele-
ment, the dentary bone, while the “post-dentary” elements,
the articular, prearticular, surangular and angular, were re-
duced to tiny, delicate ossicles. The quadrate, quadratojugal
and stapes, which were suspended from the cranium through-
out this history, were also reduced. Over roughly 100 million
years of pre-mammalian history, the bones of the auditory
chain were gradually reduced while the dentary took on a
correspondingly increased structural role in the mandible.
Biomechanical analyses describe the reduction of the bones
in the auditory chain as a sound transduction mechanism
increasingly sensitive to high frequencies (Allin, 1975, 1986;
Bramble, 1978; Crompton and Parker, 1978; Kemp, 1982;
Kermack and Kermack, 1984). A host of intricate oro-
pharyngeal functions unique to mammals probably arose con-
currently (Smith, 1992; Crompton and Hylander, 1986). Mor-
ganucodon is a transitional form in that its middle ear ossicles
morphologically resemble and probably functioned much like

76
Mammalia
ice malleus
stapes
f. vestibuli
ectotympanic
Didelphis
p ey
stapes
f. vestibuli a CMJ
quadrate a a
Ve. dentary
quadratojugal Nh
Thrinaxodon*
articular

reflected lamina
of angular
(= ectotympanic)

Probainognathus*

TIMOTHY ROWE

f. vestibuli |

stapes dentary

Mammalia

Morganucodon*

FiGURE 3. Major stages in the evolution of the mammalian mandibular arch. The angular (= ectotympanic) is shaded red, the articular
= malleus) is in green, the quadrate (= incus) is in light blue and the quadratojugal 1s in dark blue (after Rowe, 1996).

the mammalian ossicles (Allin, 1975), but they remained at-
tached to the mandible where they articulated into a narrow
groove along the medial edge of the condylar process of the
dentary and hung suspended beneath the dentary. The quad-
rate and articular also persisted as structural elements in the
CMJ (Crompton and Hylander, 1986). In pre-mammalian sy-
napsids, mastication and hearing were never fully decoupled.

This situation is transformed in mammals, in which the
postdentary bones are separated from the mandible in adults.
In addition, the entire auditory chain is displaced to a new
position entirely behind the mandibular condyle where it is
suspended solely by the adult cranium (Fig. 3). The dentary
alone forms the adult mandible, and together the dentary and
squamosal form the entire CMJ (Kemp, 1983; Gauthier et
al., 1988, 1989; Rowe, 1988, 1993; Wible, 1991). The origin
of mammals coincided with the shift from a mandibular ear
to a cranial ear as the auditory and masticatory systems be-
came decoupled. In many mammals the ear ossicles are
widely separated from the new CMJ and he behind inter-
vening secondary auditory structures such as a tympanic re-

cess or bulla, which are derived features within Mammalia
(Rowe, 1988, 1993). In their new position, the quadrate
(= incus) remains attached proximally to the stapes and dis-
tally to the articular (= malleus), while the prearticular (= os
goniale) and surangular (= ossiculum accessorium mallet) are
tightly bound or fused to the articular, and the articular is
ligamentously attached to the angular (= ectotympanic or
tympanic) which supports the tympanum. The pre-mammal-
ian linkages between the postdentary elements of the auditory
chain are thus largely conserved. The major difference is
that the quadratojugal fails to ossify and is represented, if
present at all, by a thin ligament. Apart from becoming sepa-
rated from the dentary and repositioned behind it, the mam-
malian cranial ear probably functions much as did_ the
mandibular ear of Morganucodon (Allin, 1975, 1986).
Biomechanical models elegantly explain the pre-mammal-
ian evolutionary reduction of the ear ossicles as a function
of hearing and integrated compensatory change in the
mandible (Allin, 1975, 1986; Bramble, 1978; Crompton
and Parker, 1978). But these models fail to predict or

BRAIN HETEROCHRONY

even explain detachment and repositioning of the auditory
chain, admitting that the function of the auditory chain was
probably notsignificantly altered by its detachment fromthe
mandible. Some other mechanism must be involved.

Associated Characters

The phylogenetic analysis of mammals and their extinct
relatives provided a suite of additional synapomorphies that
diagnose Mammalia and that arose at the same time the audi-
tory chain was displaced from the dentary. The association
is complex, involving the reduction and loss of bones that
were present in Morganucodon and more distant out-groups,
as well as fusions between elements that primitively re-
mained separate throughout life. Hypertrophy and heterotopy
occurred in other elements, and structures that presumably
were primitively cartilaginous later became ossified. Never-
theless, their phylogenetic association raises the possibility
that some or all of the transformations occurring in Mam-
malia ancestrally shared a common morphogenetic origin.

In the skull, the pterygoid transverse process and paroc-
cipital process were both reduced in size. The quadratojugal
and tabular were lost, as were the proatlas, atlantal rib, and
axial prezygapophysis in the neck. The squamosal became
hypertrophied to form the entire roof of the glenoid fossa.
Also hypertrophied are the occipital condyles, which became
extended upwards to enclose roughly two-thirds of the fo-
ramen magnum. The distal end of Reichert’s cartilage be-
came fused to the otic capsule where it ossifies to form the
adult mammalian styloid process. Other fusions occurred be-
tween the atlantal intercentrum and neural arches to form
the distinctive ring-like mammalian atlas. Between these
modifications of the atlas and those of the occipital condyles,
the mammalian craniovertebral joint was substantially redes-
igned. The cribriform plate was ossified, and the maxillary
turbinates became ossified as well. In addition, secondary
ossifications appeared on the limbs and girdles. More detailed
discussions of these and other characters are presented else-
where (Rowe, 1988, 1993; Gauthier et al., 1988; Wible,
1991).

While there is no obvious pattern linking all of these struc-
tures, a large number of them cluster around the brain and
lie in the same degree of proximity to the brain as the middle
ear ossicles. The influence of an inflated brain was suggested
earlier as a dominant morphogenetic influence in shaping
the unique features of the mammalian skull (Rowe, 1988,
1993). The nature of this influence can be seen more clearly
by comparing the pattern of skeletal change with a common
pattern found in the development and phylogeny of the brain.

Ontogeny and Phylogeny of the Mammalian Brain

A large brain of unique design is one of the most charac-
teristic features of extant mammals (Fig. 4). The central re-
gion of the forebrain, the telencephalic pallium, differentiates
in a singular pattern to form the isocortex (neocortex) and
pyriform cortex (Northcutt, 1984; Ulinski, 1986; Reiner,
1991: Butler, 1994). The mature isocortex forms two inflated

hemispherical lobes linked by a well-developed dorsal com-
misure. Each hemisphere has a columnar organization of six
radial layers that are generated in ontogeny by waves of
migrating cells which originate from the ventricular zone
and move radially outwards (Rakic, 1974, 1988; Walsh and
Cepko, 1992) and tangentially (Tan and Breen, 1993) to
achieve their adult positions. This inside-out pattern of neural
development is unique to mammals (Butler, 1994) and is
responsible for much of their comparatively huge cortical
volume. The mammalian cerebellum is also large in com-
parison to that of other vertebrates, with an extensively in-
folded surface and a distinct central lobe or vermis (Edinger,
1964; Jerison, 1973; Gauthier et al., 1988). For convenience,
I refer to these features collectively as an “inflated” brain.
The cerebellum follows a different developmental pattern
than does the cortex, but the cortex and cerebellum share a
common history in that an episode of expansion in both re-
gions occurred simultaneously with the detachment of the
ossicular chain.

The fossil record of extinct synapsids reveals several suc-
cessive episodes of cerebral inflation (Fig. 5). During early
synapsid history, the primitive tetrapod condition obtained
in which the brain failed to fill the adult endocranial cavity.
There is evidence in the orbitosphenoid bone of basal sy-
napsids and basal therapsids that the olfactory bulb was sus-
pended at the rostral end of a long thin peduncle which trans-
mitted the olfactory tract (Romer, 1940; Cluver, 1971). Apart
from this, few details of brain structure are preserved (Jer-
son, 1973; Hopson, 1979; Ulinski, 1986).

The basal cynodonts Procynosuchus (Kemp, 1979, 1980)
and Thrinaxodon (Hopson, 1979; Rowe et al., 1993), from
the Late Permian and Early Triassic, respectively, are the
first synapsids in which the brain filled the adult endocranial
cavity. Information about the external morphology of the
brain is preserved in these taxa in the form of natural endo-
casts and in the impressions left by the brain on the inner
surfaces of the bones that enclose it. The data sets generated
using X-ray tomography (Rowe et al., 1993, 1995) were es-
pecially informative in interpreting bone morphology with
respect to the structure of the brain (Figs. 6-8). The olfactory
bulbs appear as a slight swelling at the rostral end of the
forebrain. This reflects a second step toward the mammalian
condition in that the olfactory tracts have evidently become
engulfed from behind by the cortex, so that the olfactory
peduncle and external expression of the olfactory tract are
absent, as in mammals. At this stage, however, the circular
sulcus, which topographically demarcates the olfactory bulb
and cortex in mammalian brains (Figs. 4, 5), is not yet re-
flected in either endocasts or the bones that lie adjacent to
these structures. The forebrain was narrow, undivided, and
tubular with broad dorsal midbrain exposure between the
cerebrum and cerebellum. A long, narrow pineal foramen
(Fig. 6D) indicates the persistence of a pineal eye. Compari-
son of the cross-sectional anatomy of Thrinaxodon and
Monodelphis in coronal (Fig. 6) and transverse (horizontal)
CT imagery (Fig. 7) provides a graphic view of the extent
to which the brain expanded during the subsequent descent
of mammals.

(ZLI ‘sanoD ut) ueWIA AA PUR (1 Z6]1) PSU], UO paseg “sMarA (>) [RAIMA pur “(g) [BSIOP “(Y) FeIAIL] Yo] UL PUPIT.A sydjapiq Jo ureAG dy) JO WeABeIG] “p AANOL]

SILLIOA

TIMOTHY ROWE

Apog piozadey . - WN|[AGe1ad

aqo] sejnazoyesed ago] sejnao0yesed

suod

Apog AlesiWiwiew

smABb ajejuep

Xeon WuoyLAd
sisAydodAy ~

AINSSY [BUIYJ [eUaA ~~

gja1aqgn} AlOJOBsJO ~
BINSSI, JE|NDNO

ging A1ojae40

Wwinj9a} UIeJGpIL

XOLOIOS!
SNIINS UPIPAW

—— JINSSI] ASIOASUBI]

ging A1oj9e4/0

XO}JOI OINSSI

eae ULOJUAd /eulys

Jejnovoyesed auNSSi{

jeulysopua

pi09
jeuids qing
AIOJOBSO
wIn||aga1a0
Oren ainssif
X@LI0I0S]
wuinjoo} one JENIN

ulesgpilu OJINSSIf
OSIDASUBI]

BRAIN HETEROCHRONY

“suouttoads yuaoay WOT, ape sysvdoOpuUa xXaye] LOI, APOIIIP poyensniy sam s7ydjapiq pure
SNYIUAYAOYIUAC) “(LZH[) UOSAUIS Ade St JSRIOPUA UVOPOUOILAT (TRH ‘[e I® YORULIAY) Souog [ejaued-oyuoA UO suOIssodUU WOOL pareysn{[! eoopus
BSOIINC) Jaye SI seoopud UOJAdMAYOLIAYT (ORG) PBOUNC) LL SI ISVIOPUD SNYJDUBOUINYO.ld (EHH] “TEI OMOY) UOpoxnULAY] JO Alaseull ile) Ati-X UOINOSal YSIY pure (6/61)
uosdoy] Aq payexisni|i sJsv9opua UO paseq SI UOpOXPULIY] “(MIIADI UL “IMOY WO}) SJUOPOUAD Jo Auaso]Ayd kv oyUO paddeut ‘Mair [esiop ut s}svoopua juopousy “¢ qUNDly

eluopouAg

eluopouAong

SOUWJOJEIEWWPL\
elEWUWe |
eYdIOWINEY |

eUOUL

=
-S
~
Se
Ee
=

BRAIN HETEROCHRONY

FIGURE 7

re)

Transverse X-ray CT sections through Monodelphis domestica (A-D) and Thrinaxodon liorhinus (E-H), Sections A and E transect

the floor of the braincase: sections B and F transect the fenestra vestibuli; sections C and G transect the middle of the foramen magnum; and

sections D and H transect the roof of the foramen magnum. From Rowe (1996). See List of Abbreviations for key

A second episode of cerebral inflation is recorded in an
endocast of the Middle Triassic cynodont Probainognathus
(Quiroga, 1980). The endocast of Probainognathus is for the
first time “brain-like” (Jerison, 1973) and has begun to leave
deep impressions of its outer surface in the walls of the
osteocranium. There is now a median sulcus marking the
division between right and left olfactory bulbs and dividing
the forebrain into two incipient cerebral hemispheres (Fig
5). At this stage the “hemispheres” remain more tubular than
hemispheric, but cortical volume is relatively larger than in
Thrinaxodon. The pineal foramen is closed and the pineal
eye lost. The midbrain remains exposed dorsally, but it is
sunken between the enlarged forebrain and cerebellum

A somewhat more inflated brain is reported in the taxon
stemming from the last common ancestor of mammals and
tritheledontids, on the basis of fossils from Early Jurassic
sediments (Rowe, 1993). Therioherpeton (Quiroga, 1984), a
poorly known basal member of this group (Fig. 5), has a
brain-like endocast reportedly larger than Probainognathus

(Quiroga, 1980) but no newly differentiated features are dis-
cernible. Scaling may introduce an element of artifact into
the perception of a larger brain, for the basal members of
this clade are much smaller than Probainognathus and more

distant cynodonts. Other early Jurassic fossils indicate that

further inflation occurred in the taxon stemming from the
last common ancestor of Sinoconodon and mammals. This
is suggested by such features as bulging of the parietals out-
ward into the temporal fenestra and bony flooring beneath
the cavum epipterycum (Crompton and Luo, 1993; Rowe

1993). The inner surfaces of the parietal-interparietal of Si
Edinger, 1964; Jer-

ison, 1973) and Morganucodon (Kermack et al., 1981) pre

noconodon (Patterson and Olson, 1961

serve impressions left by the divergent caudal poles of the
forebrain, which span a wider curvature than in Thertohe)
peton. Like those of more plesiomorphic cynodonts, how
ever, the olfactory bulbs remained almost cylindrical and
lacked any topographic demarcation from the cerebrum
Additional plesiomorphic features include confinement of the

ue Sauog (1
O19YM “(7A

SS
oa
~
~
2

BRAIN HETEROCHRONY

cerebral hemispheres to a narrow space between the ascend-
ing processes of the epipterygoids (alisphenoids), persistent
exposure of the midbrain dorsally, and the front of the brain-
case remaining unenclosed.

The next major episode of cerebral expansion is recorded
in features shared by all extant mammals and which are most
parsimoniously interpreted as having arisen in their last com-
mon ancestor. The olfactory bulbs are inflated, hemispheri-
cal, and sharply differentiated in their external morphology
from the cortex by the circular fissure, which now is visible
on endocasts for the first time (Fig. 5). The forebrain is
greatly inflated into two hemispherical lobes that expand
backwards to cover the midbrain. It appears that this is the
point in history at which the cortex differentiated into a sepa-
rate isocortex and pyriform cortex. The isocortex, present in
all mammals, is recognizable in endocasts of relatively primi-
tive mammals by its degree of inflation, convex hemispheri-
cal form, and backward expansion over the midbrain. In ex-
ternal morphology of the brain itself, the isocortex is also
distinguished by both the circular fissure and rhinal fissure.
The rhinal fissure classically has been used to identify the
isocortex boundaries in endocasts of placentals (Jerison,
1973, 1990). However, it is rarely discernible in endocasts
of the small, primitive taxa of relevance here. For example,
the rhinal fissure is visible on the surface of didelphid brains
(Ulinski, 1971) but it does not appear in their endocasts. The
circular fissure is generally discernible in small endocasts
and is the more diagnostic of the two fissures among the
taxa of interest. An additional innovation of the mammalian
brain is that the cerebellum is inflated and deeply folded,
with a distinct vermis projecting rostrally along the midline
between the caudal ends of the cerebral hemispheres. These
cerebellar features are not discernible in the bones of the
braincase of Morganucodon and more distant relatives.

The oldest fossils preserving skeletal apomorphies derived
within Mammalia are from Middle Jurassic sediments (Rowe,
1988, 1993), and an endocast displaying the mammalian fea-
tures described above 1s preserved in the Late Jurassic 7ri-
conodon mordax (Simpson, 1927). In Triconodon the man-
dible is comprised solely of the dentary, and its enclosed
Meckelian sulcus indicates that the ossicular chain had be-
come detached from the jaw. The preserved features in this
endocast are similar to didelphid endocasts in shape and rela-
tive size (Fig. 5). In later mammalian history the rate of
brain evolution varied remarkably among the lineages that
have survived until today, with didelphids reflecting the least
subsequent evolution and cetaceans and primates showing
the greatest. Cerebral inflation in mammals is widely held
to have evolved in relation to the invasion of a nocturnal
and perhaps arboreal niche. Cortical expansion and differ-
entiation into 1socortex and pyriform cortex support height-
ened olfactory and auditory senses (Jerison, 1973), and co-
incident, overlapping sensory and motor maps of the entire
body surface (Lende, 1963a, b, c). Cortical expansion has
also been implicated in the evolution of endothermy (Jerison,
1973; Allman, 1990). The enlarged cerebellum is related to
the adaptive coordination of movement through a complex
three-dimensional environment (Thach et al., 1992).

The origin of the inflated brain in mammals reflects an
episode of heterochrony in which the brain began to grow
both faster and longer into ontogeny than it did in non-mam-
malian cynodonts. This is clearly an instance of peramor-
phosis, where the descendant ontogeny transcends the ter-
minal state achieved during development by its ancestors
(Gould, 1977; Alberch et al., 1979; Fink, 1982; Kluge, 1988).
Without more knowledge about the relative timings and
growth rates of developmental trajectories in the extinct out-
groups, it 1s not possible to discern what type of peramor-
phosis (hypermorphosis, acceleration, predisplacement) has
occurred. In the absence of direct experimental evidence, the
most likely genetic moderation of this event now appears to
lie in the homeobox genes and homeodomain proteins which
direct early patterning in vertebrates generally (Rakic, 1988;
Wilkinson et al., 1989; Keynes and Lumsden, 1990; Gilbert,
199]; Langille and Hall, 1993; Rubenstein et al., 1994; Hol-
land, 1996:63—70). In the developing hindbrain, homeobox
genes control the identity of rhombomeres, which are seg-
mental bulges that confine clones of cells and domains of
differential gene expression (Walsh and Cepko, 1992). Fore-
brain segmental patterning is now known to be under a simi-
lar control (Rubenstein et al., 1994). Simply specifying more
segments during early pattern formation may produce an en-
larged adult brain, although there is as yet no experimental
verification (Marx, 1992). Whatever the genetic control, it
is evident that a heterochronic perturbation of the central
nervous system occurred in mammals ancestrally, producing
differential growth of the brain that launched a cascade of
secondary, epigenetic effects.

Epigenetic Influences on Cranial Development

It is well established that familiar physical forces and dy-
namic processes are significant mechanisms in pattern for-
mation and morphogenesis throughout ontogeny (Oster et
al., 1985, 1988; Newman and Comper, 1990). These forces
and processes include, among others, gravity (Malacinski,
1984), adhesion (McClay and Ettensohn, 1987; Armstrong,
1989), diffusion (Crick, 1970), interfacial tension (Steinberg,
1978; Heintzelman et al., 1978), mechanical loading (Hoyte,
1966, 1975; Moss, 1968; Hall, 1984a,b,c, 1992; Wong and
Carter, 1990; Herring, 1993a,b), electrical potentials (Bassett,
1972; Metcalf and Borgens, 1994; Metcalf et al., 1994), ma-
ternal biological rhythms (Lloyd and Rossi, 1993), viscous
flow, phase separation, Marangoni effects, convective fin-
gering, chemical concentrations, and density (Newman and
Comper, 1990). Newman and Comper (1990) argued that
morphogenic and patterning effects are the inevitable out-
come of these recognized physical properties of cells and
tissues. Many of these forces and processes can affect skele-
togenesis, and there is ample observation and experimenta-
tion to indicate that the skeleton 1s responsive to a hierarchy
of such influences from the time of earliest condensation of
proskeletal tissues through old age (Wong and Carter, 1990).

Newman and Comper (1990) refer to these mechanisms
as “generic” physical processes, while others (e.g., Hall,

1990, 1992; Herring, 1993a,b) refer to them under the more

inclusive term “epigenesis.” These forces may complement
and act in concert with biomolecular (genetic) processes, or
they may operate by themselves, or not at all in any particular
developmental episode. When invoked, they may have broad
spatial effects that touch different populations of cells and
different tissue types. Many of these processes are known
to have nonlinear responses to relevant control variables,
such that small changes in rate or magnitude of a process,
or through limited interaction between parts can lead to pro-
found effects in the resulting morphology (Mittenthal, 1989).
Hall (1990, 1992) refers to this as the spatial and temporal
cascading effect of ontogeny, which can produce new and
unexpected consequences for adult structure. Major morpho-
logical reorganizations in phylogenetic lineages may arise
by the action of these mechanisms at different times in on-
togeny. The effects potentially are more profound as the
forces act during earlier stages in development.

A vast medical, anatomical, experimental, and theoretical
literature describes the response of postnatal cramiofacial
growth in humans and other placental mammals to mechani-
cal loading (e.g., D’Arcy Thompson, 1942; Huber, 1957;
Moss, 1958, 1968; Hoyte, 1966 1971, 1975; Bassett, 1972;
Pritchard, 1972; van Limborgh, 1972; Buckland-Wright,
1978; Spyropoulos, 1978: Babler and Persing, 1982; Hurov,
1986; Storey and Feik, 1986; Carter, 1987; Carter and Wong,
1988; Wong and Carter, 1990; Herring, 1993a,b). In the ear-
liest stages of skeletal development, mechanical loading ts
probably far less important to basic patterning than cell-to-
cell adhesion, surface tensions, chemical gradients, and other
epigenetic forces that act primarily at molecular and cellular
levels. But from the time that tissues are differentiated and
individual organs begin to grow, a new level in the epigenetic
hierarchy may be expressed as loads are generated by dif-
ferential growth.

Growth and form of the skull reflect the dynamic interac-
tion of structural elements and epigenetic forces throughout
ontogeny. Through much of organogenesis and early growth,
the most significant forces are generated by expansion of
the brain and its special sense organs, especially the eye.
That the embryonic brain actually loads surrounding tissues
is evident in the nature of its growth. Brain enlargement in
early ontogeny 1s driven by a combination of tissue growth
and hydrostatic volume increase in the medullary cavity. Fol-
lowing neurulation, the tubular brain becomes a hydrostatic
reservoir as the rostral neuropore closes and the spinal neuro-
coel becomes occluded and the medullary cavity between
them fills with an increasing volume fluid. Proper intraven-
tricular pressure 1s required to drive brain expansion (Jelinek
and Pexieder, 1968; 1970a, b: Desmond and Jacobson, 1977;
Goodrum and Jacobson, 1981; Pacheco et al., 1986). The
law of LaPlace describes the distending tension in the wall
of a cylindrical vessel at any given pressure as directly pro-
portional to the vessel’s radius (Gardner, 1973; Pacheco et
al., 1986). The volume of the medullary cavity increases at
a linear rate while brain tissue growth increases exponen-
tially, in part as a mechanical requirement to prevent the
brain from bursting as its outer tension rises. Cerebral loading
onto surrounding tissues is thus proportionate to the sum of

TIMOTHY ROWE

hydrostatic load plus the load from the growing cerebral tis-
sue. Severe deformities of the skull accompany pathologies
such as microcephaly and anacephaly, which result from dis-
ruptions in ventricular pressure during early development
(D’Abundo, 1905; Weed, 1920; Nanagas, 1925; Young,
1959; Hoyte, 1966; Moss, 1968; Gardner, 1973; Herring,
1993a).

By the time the first skeletal condensations appear in mam-
mals, the tissues in which they differentiate are already
stretched around a cylinder that is relatively larger than that
occurring even in the terminal stages of ontogeny of the
closest extinct relatives of mammals. As can be seen in the
comparative CT sections of Monodelphis and Thrinaxodon
(Figs. 6, 7), the mammalian bones span cerebral surfaces of
greater curvature and are correspondingly thinned, suggesting
that the materials to construct the skull did not increase at
the same accelerated rate of growth as the brain. In Monodel-
phis the cranium is largely enclosed by bone in the fourth
week but the brain continues to grow through the twelfth
week and the skeleton is continually remodeled throughout
the intervening period (Fig. 8). Both experimental and tera-
tological evidence indicate that cerebral loading affects skele-
tal growth from the very beginnings of mesenchymal con-
densation, through chondrogenesis, and for a considerable
portion of skeletal growth.

In addition to influencing connective tissue growth, me-
chanical loads can direct cell differentiation. An outstanding
example is the adaptive and compensatory responsiveness of
mammalian secondary cartilage and intramembranous bone
to loading in the mechanical environment created during the
repair of bone fractures, an ability that is expressed early in
ontogeny and which persists into adult life. For example,
along angulated fractures in broken limb bones, first chon-
drogenesis and then endochondral ossification are induced
by compressive loads on the concave side, while intramem-
branous ossification commences on the convex side of a re-
pairing shaft (Pritchard, 1972; Hall, 1975, 1984a, b, c, 1992:
Herring, 1993b; Wong and Carter, 1990). Another such
modulation is the condylar secondary cartilage of the mam-
mahan dentary. Loading initiates the differentiation of sec-
ondary cartilage in cells that can differentiate either as chon-
droblasts or osteoblasts. Reduction of condylar loading
suppresses secondary chondrogenesis and initiates intramem-
branous ossification (Hall, 1984a, 1992; Herring, 1993b).

The developing cranial muscles may generate loads of
comparable magnitude to those of the developing brain as
they grow and begin to twitch and contract. and they have
been implicated in the detachment of the auditory chain (Her-
ring, 1993a: Maier, 1987). Experimental data indicate that
embryonic muscular movement not only loads the skeleton,
but that these loads are critical to the proper differentiation
of joints and joint capsules (Drachman and Sokoloff, 1966;
Murray and Drachman, 1969; Laing, 1982). As muscles ap-
proach maturity they become capable of exerting far greater
levels of load than the growing brain or developing
myoblastemata. Muscular loading induces the mature form
of such features as the coronoid and angular processes of
the mandible, it contributes significantly to shaping the ma-

BRAIN HETEROCHRONY

ture craniomandibular and craniovertebral joints, and to
growth of the lambdoidal and sagittal crests (e.g., Hoyte,
1966, 1971, 1975; Spyropolous, 1978; Hurov, 1986; Carter,
1987; Carter and Wong, 1988). In generating these extreme
levels of force, muscular loading can induce a new level in
the hierarchy of epigenesis which may be expressed long
into ontogeny after the effects of differential growth are spent
(Fig. 8).

When Maier (1987) and Herring (1993a) implicated mus-
cle loading in the detachment of the auditory chain, they
followed earlier authors (e.g., Allin, 1975; Crompton and
Parker, 1978) in supposing that mammalian ontogeny reca-
pitulates the transformation between two functional joints,
that is from a functional primary CMJ between the palato-
quadrate and articular cartilages to the mature CMJ between
the dentary and squamosal bones. More recent research sug-
gests that this is not the case. In a histological study of the
developing CMJ in Monodelphis, Filan (1991) found no evi-
dence to suggest a functional joint ever forms between the
quadrate and articular cartilages before they become detached
and the dentary-squamosal joint becomes functional. In cap-
tivity, the young do not begin eating solid food until they
are 4 to 5 weeks old (Fadem et al., 1982; Kraus and Fadem,
1987), following detachment. Secondary condylar cartilage
and the beginnings of a synovial capsule also appear during
the fourth week at the joint between the dentary and
squamosal and signal the onset of CMJ loading by the mas-
ticatory muscles that insert on the dentary. It 1s difficult to
precisely define a time at which the dentary-squamosal joint
becomes functional, because for a time the contacts between
condylar cartilage and the squamosal and the auditory ossi-
cles and the otic capsule are equally large (Clark and Smith,
1993). As ontogeny progresses, the masticatory muscles
transmit increasing loads to the CMJ and correspondingly
its surface increases, mostly through a process of lateral ac-
cretion as the width between the right and left CMJs increases
(Fig. 12).

Muscular loading fails to completely explain the develop-
mental transformation of the ear ossicles in mammals. While

muscular loading might contribute to early differentiation of

the mandibular and auditory elements and to the initial tear-
ing of the connective tissues that bind the ossicular chain to
the mandible, this interpretation is complicated by the timing
of the event in different mammals. In marsupials it takes
place after birth and the young have begun to suckle, while
in placentals it occurs before birth, making it difficult to
identify a common mechanical setting. More importantly,
the force trajectories of the masticatory muscles are oriented
in such a way that the mandibular condyle is pulled upwards
and backwards into the roof of the glenoid, compressively
loading the craniomandibular joint (Crompton and Hylander,
1986). It is difficult to see how this action could lead to the
posterior repositioning of the auditory chain behind the den-
tary condyle; masticatory loading would be more likely to
press the dentary backwards against the postdentary bones
than to separate the two. If the masticatory musculature is
involved at all, its role is only part of the story and some

other mechanism must be responsible for widely separating
the auditory chain from the mandible.

Development of the Middle Ear Ossicles

The developing auditory chain has both endochondral and
intramembranous components, and both types have attach-
ments to the mandible that are broken as ontogeny pro-
gresses. Three cartilages are present at birth in Monodelphis.
The stapes has already budded from Reichert’s cartilage and
forms a tiny rod with a small footplate that lies in the center
of the opening of the fenestra vestibuli. Both the stapes and
the petrosal eventually contribute to the formation of the
mature footplate in later in ontogeny as a complex stapedial
articulation develops at the fenestra vestibuli. Articulating
with the distal end of the stapes is the caudal moiety of the
palatoquadrate cartilage, which is braced against the ventro-
lateral edge of the otic capsule and which will ossify to form
the incus (= quadrate). Meckel’s cartilage forms a continuous
elongate rod that bends downward at its rear end at nearly
a right angle (Fig. 9). During the second week, the rear ex-
tremity is cleaved from the mandibular ramus of Meckel’s
cartilage, forming the cartilage in which the malleus (= ar-
ticular) ossifies. The two pieces become separated when
Meckel’s cartilage degenerates and is resorbed during ossi-
fication of the dentary.

The intramembranous ossifications have a contrasting de-
velopmental history. At birth, both the dentary and ectotym-
panic (= angular) have begun to ossify in a common mem-
branous sheet external to Meckel’s cartilage, but at this stage
their growth centers are widely separated and an expanse of
connective tissue intervenes (Fig. 9). During the first three
postnatal weeks, the ectotympanic grows in positive al-
lometry relative to the dentary. As the ectotympanic grows,
it expands against the developing angular and condylar proc-
esses of the dentary, and the two bones are held together by
fibrous connective tissue that arises in the osteogenic mem-
brane. During early ontogeny the ectotympanic lies in its
ancestral position hanging beneath the condylar process of
the dentary. By the end of the third week the ectotympanic
is approaching its adult size. At this time its growth rate
slows and shifts into a negative allometry that persists for
the remainder of ontogeny. At roughly this same time, the
ectotympanic is torn free from the dentary (McClain, 1939;
Clark and Smith, 1993). During the next 9 weeks the auditory
chain migrates backwards from beneath the condylar process
and eventually comes to rest entirely behind and medial to
craniomandibular joint (Figs. 3, 9).

The key to understanding both the detachment and sub-
sequent relocation of the auditory chain may lie in an inter-
play between the differential growth among elements of the
mandibular arcade and the brain. The brain balloons upwards
and backwards from the developing facial skeleton and grows
steadily for the first 12 weeks (Fig. 10) of postnatal ontogeny
(Ulinski, 1971). The relative positions of the CMJ and fenes-
tra vestibuli are convenient markers to follow in tracing cra-
nial remodeling in the wake of cerebral growth (Figs. 11,

86 TIMOTHY ROWE
drawn to same length drawn to same scale

adult

day 90 day 81

day 60
day 55

day 30 day 34

Sica / day 21

( > day6

OQ day 1

—
4mm
Figure 9. Left Development of the mandibular arch in Monodelphis domestica, based on video imagery of a cleared and double-stained

erowth series, drawn to same length, Cartilage is shaded blue and the membranous ectotympanie is in red. Right: Growth of the forebrain in
Did Iphis, from birth to adult (modified after Ulinski, 1971), drawn to same scale. Based on Rowe (1996)

BRAIN HETEROCHRONY

.30.

Ectotympanic
Diameter
Dentary Length 10

~<— time of detachment

Dentary Length

Ectotympanic

Diameter

Forebrain Length

: yy

22mm +
18 4
Forebrain 1a4
Length 10 +
6 +
2mm
; ; 3.07
Posterior Height
Anterior Height .
(A H) 2.04 . . ery Les
2.07
Cortical Length
Posterior Height a . -
(P H) 1.04 oo
Cortical Length
Occipital Length
(OL)

10 20 30 40 50

Ht
60 70 80

adult

FiGure 10. Differential growth of the forebrain (based on Didelphis, after Ulinski, 1971) and ectotympanic (based on Monodelphis). In the
top graph, growth of the diameter of the ectotympanic is plotted as a function of the length of the mandible. In the second graph, growth of the
total combined length of the olfacctory bulb plus cortex 1s plotted as a function of age. In the lower three graphs, relative growth of the different
parts of the cortex is plotted as a series of ratios defined by the dimensions depicted on the mature brain. The ratios show a phase of ventro
caudal growth (days 0-10), a phase of anterior growth (days | 1-40), and a phase of occipital growth (days 41-81). From Rowe (1996)

12). At birth, the fenestra vestibuli lies immediately behind
and medial to the CMJ, in a relationship similar to that found
in adult Morganucodon and Thrinaxodon. The fenestra
vestibuli and CMJ both lie external to the developing cerebral
vesicle, along roughly the same “latitude” of cerebral cir-
cumference, which I refer to informally as the cortical “equa-
tor” (Fig. 11). As the brain grows, the magnitude of curvature
of equator grows as well. The distance between the fenestra
vestubuli and the CMJ, which both lie on the equator, also
increases. The entire rear part of the skull appears to be
pushed backwards from the facial skeleton and mandible by
the growing brain.

The equatorial segment between the fenestra vestibuli and
the CMJ defines an arc of detachment (Fig. 12) whose mag-
nitude of curvature expands as the brain grows. As curvature
of the arc expands, the fenestra 1s displaced progressively
backwards. For about the first three weeks, the ear ossicles
grow at a sufficient pace to keep up with the growing arc,
thus maintaining their primitive linkage between the fenestra
vestibuli and the mandible. As the ossicular growth rate slows
and shifts into negative allometry, the brain continues its
pace of growth for nine additional weeks and undergoes a
ten-fold increase in volume during that time (Ulinski, 1971).

During this time the arc nearly doubles in curvature, bursting
the primitive arcade of skeletal elements that had spanned
from the mandibular symphysis to the fenestra vestibuli. The
middle ear bones maintain their attachment to the fenestra
vestibuli and follow its trajectory backwards from the time
of their detachment at the end of the third week until the
brain stops growing in the twelfth week.

The timing of detachment prevents the disruption of func-
tion in the middle ear bones because it occurs before the
onset of auditory functionality. The ear is unresponsive to
sound until the 6th week and only thereafter does the auditory
tract become myelinated (Langworthy, 1928; Larsell et al.,
1935; McCrady et al., 1937; McCrady, 1938; McClain,
1939). The geometry of the widening arc of detachment ac-
counts for the detachment of the auditory chain, for the pre-
cise path of its subsequent posterior displacement, and for
the timing and extent of this movement in both ontogeny
and phylogeny.

Discussion

The phylogenetic concordance of the inflated brain and
the cranial ear implied the unexpected possibility of a causal

ROWE!

>
=

IINQIISSA BNSBUS}

Joyenbs
JBOD

BRAIN HETEROCHRONY

plane of cortical equator

adult

day 90

day 60
day 27

day 0

plane of cortical equator

FIGURE 12. A) Location ofa 1 mm-thick slice crossing the cerebral equator, fenestra vestibuli, and glenoid across the skull of a didelphid
B) Growth of the forebrain showing location of cortical equator. C) Superimposed cerebral equators of Monodelphis from day 0, day 27, day
60, day 90 and adult, relating the growth trajectories of the cerebral equator, fenestra vestibuli, and CMJ. Equatorial profiles taken from high

resolution X-ray CT imagery of Monodelphis. From Rowe (1996).

relationship between the two structures, and this relationship
appears to be corroborated by their ontogeny. The negative
allometry of the auditory chain in the wake of continued
rapid growth of the brain combine to cause the auditory chain
to be detached from the mandible and carried backwards to
its mature position behind the mandible. This new relation-
ship originated in mammals ancestrally in an episode of het-
erochronic increase in rate and duration of brain growth. This
one mechanism appears primarily responsible for both the
evolutionary origin of the mammalian middle ear and its
recapitulation in ontogeny. If this interpretation is correct,
an event of fundamental importance in the origin of mammals
was a heterochronic perturbation of brain development. As
the pace and duration of brain development reached the an-
cestral mammalian level, a cascade of secondary, epigenetic
effects was unleashed that affected virtually all aspects of
mammalian life history.

One class of cascading effects involves intrinsic features
of the brain and the many functions it controls. The
specification of mammalian cortical regions is largely epige-

netic as it occurs following neurogenesis, while clones of
cortical neurons mingle during subsequent development.
Neurogenesis appears to produce a cortex that is initially
uniform and that later differentiates into specific functional
areas by intercellular interactions (Walsh and Cepko, 1992),
a process occurring over a protracted period of postnatal on-
togeny. In the newborn opossum, for example, the cortex is
unlayered, and subsequent development of its external ap-
pearance over the next 10 weeks mirrors many aspects of
histogenesis and architectonic differentiation occurring at the
same time (Riese, 1945; Ulinski, 1971). The extended dura-
tion of cerebral ontogeny that arose ancestrally in mammals
afforded the specification of many new structures and an
increased capacity for learning, both neuromuscular and as-
sociative, which continues long after cerebral differentiation
and growth have ceased. Specific changes in cortical circuitry
arising with expansion of the mammalian brain are related
predominantly to elaboration of sensory components and en-
hancement of motor control. Modality-specific sensory
channels through the thalamus to the telencephalon, which

were probably present in amniotes ancestrally, became ex-
panded in association with an extended range of auditory
frequencies, enhanced olfaction, and with the sensory func-
tion of hair (Ulinski, 1986; Butler, 1994). Also distinctively
mammalian are the development of corticospinal (palliosp1-
nal) pathways (Northcutt, 1984), and well-developed specific
motor nuclei which receive afferents from the cerebellum or
basal ganglia, project to specific restricted regions of the
cortex, and are situated rostrally in the ventral half of the
thalamus (Ulinski, 1986). Mammals are further characterized
by divided optic lobes, development of the pons varoli, and
elaboration of the inferior olive and pontine nuclei (Ulinski,
1986; Gauthier et al., 1988). These features collectively re-
sulted in elaboration of the sensorimotor system to a degree
surpassing all other vertebrates (Ulinski, 1986; Butler, 1994).
The effects of this cortical elaboration are manifested during
life history in functions ranging from the complex repertoire
of mammalian oropharyngeal functions (Smith, 1992) to the
maintenance of rhythmic respiratory movements associated
with mammalian metabolism (Carpenter, 1976) to the diverse
patterns of mammalian locomotion (Bramble, 1989; Bramble
and Jenkins, 1993). Some of these functions surely extend
into pre-mammalian history, but the marked increase in cere-
bral differentiation and volume that occurred in Mammalia
ancestrally suggests a marked increase in functionality com-
pared with the conditions in Morganucodon and more distant
synapsids.

Another class of epigenetic cascade induced alterations in
structures extrinsic to the brain, especially the adjacent con-
nective tissues. The shift to a cranial middle ear is the most
notable example, but virtually all parts of the skull and neck
near the brain were also modified. The pattern of skeletal
modification is complex, involving an interplay of reduction,
loss, fusion, hypertrophy, and heterotopy of the components.
Comparable patterns of complex change are manifested by
a variety of developmental pathologies of the skeleton which
are traceable to early perturbations of the mesenchymal tis-
sues in Which the skeleton differentiates and which can be
traced to mutations of single genes (Grtineberg, 1963).

Because heterochrony and its secondary effects are impos-
sible to identify without a phylogeny, it is not surprising that
the effects of brain heterochrony on the mammalian skeleton
were unrecognized under the phenetic Linnean view of early
mammalhan history. The assertions of convergent evolution
and the lack of obvious adult biomechanical or physiological
correlation between the middle ear and brain further obscured
the relationship of ear morphology to cerebral growth. The
discovery of this unsuspected relationship between the brain
and ear illustrates the potential value of phylogenetic sys-
tematics to the many developmental and experimental dis-
ciplines within biology which now operate largely in the
absence of a well-corroborated phylogenetic framework.
Within such a framework, experimental manipulations of de-
veloping mammals can be designed to further test the rela-
tionship between genetic and epigenetic factors in onto-
geny, and to elucidate the mechanisms of evolutionary
change in the historical context in which they evolved.

TIMOTHY ROWE
Acknowledgments

I am grateful to Drs. Michael Ghiselin and Giovanni Pinna
for inviting me to participate in this symposium, and for
providing such a productive and invigorating forum in which
to present this work. I thank Mr. Reuben Reyes for invaluable
assistance in generating and processing the exquisite digital
datasets of high resolution X-ray CT imagery used in this
study. Dr. Rafael de Sa and Ms. Hillary Tulley prepared
histological materials and cleared and stained the extensive
growth series of Monodelphis domestica used here. Mr. Chris
Brochu, Dr. David Cannatella, Mr. Matthew Colbert, Dr. Er-
nest Lundelius, Jr., Dr. Zhext Luo, and Mr. John Merck, Jr.
read earlier drafts of this manuscript, and their stimulating
and insightful discussions contributed significantly to all
phases of this research. My thanks to John Merck, Egan
Jones, and Jeffrey Horowitz, who provided some of the il-
lustrations and assisted with various aspects of generating
the imagery used herein. This research was sponsored by
National Science Foundation grants BSR-89-58092 and
USE-91-56073, University of Texas Geology Foundation,
and the Vertebrate Paleontology and Radiocarbon Labora-
tory.

Literature Cited

Alberch, P., S. J. Gould, G. F. Oster, and D. B. Wake. 1979. Size
and shape in ontogeny and phylogeny. Paleobiology 5:296-317.

Allin, E. F. 1975, Evolution of the mammalian middle ear. Journal
of Morphology 147:403-438.

— _ 1986. The auditory apparatus of advanced mammal-like
reptiles and early mammals. Pages 283-294 in N. Hotton III, P.
D. MacLean, J. J. Roth and E. C. Roth, editors. The Ecology and
Biology of Mammal-Like Reptiles. Smithsonian Institution Pres,
Washington.

Allman, J. 1990. The origin of the neocortex. The Neurosciences
2:257-262.

Armstrong, P. B. 1989. Cell sorting out: the self-assembly of tissues
in vitro. Critical Reviews in Biochemistry and Molecular Biology
24:119-149.

Babler, W. J. and J. A. Persing. 1982. Experimental alteration of
craial suture growth: effects on the neurocranium, basicranium,
and midface. Pages 333-345 in Factors and Mechanisms Influ-
encing Bone Growth. Alan R. Liss, New York.

Bassett, C. A. L. 1972. A biophysical approach to craniofacial mor-
phogenesis. Acta Morphologica Nerlando-Scandinavica 10:71]
SO.

Bramble, D. M. 1978, Origin of the mammalian feeding complex:
models and mechanisms. Paleobiology 4:271—301.

. 1989. Axial-appendicular dynamics and the integration of
breathing and gait in mammals. American Zoologist 29:171—-186,

BRAIN HETEROCHRONY

Bramble, D. M. and F. A. Jenkins Jr. 1993. Mammalian locomo-
tor-respiratory integration: implications for diaphragmic and pul-
monary design. Science 262:235—240.

Buckland-Wright, C. J. 1978. Bone structure and the patterns of
force transmission in the cat skull (Felis catus). Journal of Mor-
phology 155:35—02.

Butler, A.B. 1994. The evolution of the dorsal pallium in the
telencephalon of amniotes: cladistic analysis and a new hypothe-
sis. Brain Research Reviews 19:66-101.

Carpenter, M. B. 1976. Human Neuroanatomy, 7th Edition. Wil-
liams and Wilkins Company, Baltimore.

Carter, D. R. 1987. Mechanical loading history and skeletal biology.
Journal of Biomechanics 20:1095—1109.

Carter, D. R. and M. Wong. 1988. The role of mechanical loading
histories in the development of diarthroidal joints. Journal of Or-
thopedic Research 6:804-816.

Clark, C.T. and K.K. Smith. 1993. Cranial osteogenesis in
Monodelphis domestica (Didelphidae) and Macropus eugenii
(Macropodidae). Journal of Morphology 215:119—149.

Cluver, M. A. 1971. The cranial morphology of the dicynodont ge-
nus Lystrosaurus. Annals of the South African Museum 56:155—
274.

Coues, E. 1872. On the osteology and myology of Didelhis vir-
giniana, with an appendix on the brain by Jeffries Wyman. Mem-
oirs of the Boston Society for Natural History 2:41—154.

Crick, F. H. C. 1970. Diffusion in embryogenesis. Nature 225:420—
422.

Crompton, A. W. and W. L. Hylander. 1986. Changes in mandibular
function following the acquisition of a dentary-squamosal jaw
articulation. Pages 263-282 in N. Hotton III, P. D. MacLean, J.
J. Roth and E. C. Roth, editors. The Ecology and Biology of
Mammal-Like Reptiles. Smithsonian Institution Press,
Washington.

Crompton, A. W. and F. A. Jenkins Jr. 1973. Mammals from rep-
tiles: a review of mammalian origins. Annual Review of Earth
and Planetary Sciences 1:131—155.

Crompton, A. W. and Z. Luo. 1993. Relationships of the Liassic
mammals Sinoconodon, Morganucodon oehleri, and Dinneth-
erium. Pages 30-44 in F.S. Szalay, M. J. Novacek and M. C.
McKenna, editors. Mammalian Phylogeny. Springer Verlag, New
York.

Crompton, A. W., and Parker. 1978. Evolution of the mammalian
masticatory apparatus. American Scientist 66:192—201.

D’Abundo, G. 1905. Atrofie cerebrali sperimentali ed atrofie cra-
niensi concomitanti. Annali di Neurologia 23:397-402.

de Beer, G. 1937. The Development of the Vertebrate Skull. Claren-
don Press, Oxford. (Pl. 1-143.)

. 1958. Embryos and Ancestors, 3rd Edition. Clarendon
Press, Oxford.

de Queiroz, K., and J. A. Gauthier. 1990. Phylogeny as a central
principle in taxonomy: phylogenetic definitions of taxon names.
Systematic Zoology 39:307-322.

. 1992. Phylogenetic taxonomy. Annual Review of Ecology
and Systematics 23:449-480.

~. 1994. Toward a phylogenetic system of biological nomen-
clature. Trends in Ecology and Evolution 9:27-31.

Desmond, M. E. and A. G. Jacobson. 1977. Embryonic brain en-
largement requires cerebrospinal fluid pressure. Developmental
Biology 57:188—198.

Drachman, D. B. and L. Sokoloff. 1966. The role of movement in
embryonic joint development. Developmental Biology 14:401—
420.

Edlinger, T. 1964. Midbrain exposure and overlap in mammals.
American Zoologist 4:5—19.

Fadem, B. H., G. L. Turpin, E. Maliniak, J. L. VandeBerg, and V.
Hayssen. 1982. Care and breeding of the gray, short-tailed opos-
sum (Monodelphis domestica). Laboratory Animal Science
32:405—-409,

Farris, J. S. 1986. HENNIG &6 software, version 1.5.

Filan, S. L. 1991. Development of the middle ear region in Monodel-
phis domestica (Marsupialia, Didelphidae): marsupial solutions
to an early birth. Journal of Zoology (London) 225:577—S88.

Fink, W. L. 1982. The conceptual relationship between ontogeny
and phylogeny. Paleobiology 8:254-264.

Gardner, W. J. 1973. Hydrodynamic mechanisms. Pages 15-21 in
W. J. Gardner, editor. The Dysraphic States: Syringomyelia to
Anacephaly. Exerpta Medica, Amsterdam.

Gauthier, J. 1994. The diversification of amniotes. Pages 129-159
in D.R. Prothero and R.M. Schoch, editors. Short Courses in
Paleontology, No. 7: Major Features of Vertebrate Evolution. Pa-
laeontological Society, Knoxville

Gauthier, J., D. Cannatella, K. De Queiroz, A. G. Kluge, and T.
Rowe. 1989. Tetrapod phylogeny. Pages 337-353 in B. Fern-
holm, K. Bremer and H. Jérnvall, editors. The Hierarchy of Life:
Molecules and Morphology in Phylogenetic Analysis. Exerpta
Medica, Amsterdam.

Gauthier, J., A. G. Kluge, and T. Rowe. 1988. Amniote phylogeny
and the importance of fossils. Cladistics 4:105—209.

Gilbert, S. F. 1991. Developmental Biology, 3rd Edition. Sinauer
Associates, Sunderland.

Goodrich, E. S. 1930. Studies on the Structure and Development
of Vertebrates. Macmillan, London.

Goodrum, G.R. and A. G. Jacobson. 1981. Cephalic flexure for-
mation in the development of the chick embryo. Journal of Ex-
perimental Zoology 216:399-408.

Gould, S.J. 1977. Ontogeny and Phylogeny. Harvard University
Press, Cambridge.

Griineberg, H. 1963. The Pathology of Development: a Study of
Inherited Skeletal Disorders in Animals. John Wiley and Sons,
New York.

Hall, B. K. 1975. Evolutionary consequences of skeletal differen-
tiation. American Zoologist 15:329-350.

———. ]984a. Genetic and epigenetic control of connective tissues
in the craniofacial structures. Birth Defects: Original Article Se-
ries 20:1—17.

————. 1984b. Developmental processes underlying heterochrony
as an evolutionary mechanism. Canadian Journal of Zoology
62:1—7.

————, 1984c. Developmental processes underlying the evolution
of cartilage and bone. Symposia of the Zoological Society of
London 5§2:155—176.

———, 1990. Genetic and epigenetic control of vertebrate embry-
onic development. Netherlands Journal of Zoology 40:352-361.

——. 1992. Evolutionary Developmental Biology. Chapman and
Hall, London.

Hanken, J. 1993. Model systems versus outgroups: alternative ap-
proaches to the study of head development and evolution. Ameri-
can Zoologist 33:448-456.

Hanken, J. and B. K. Hall, editors. 1993. The Skull. University of
Chicago Press, Chicago.

Heintzelman, K. F., H. M. Phillips, and G. S. Davis. 1978. Liquid-
tissue behavior and differential cohesiveness during chick limb
budding. Journal of Embryology and Experimental Morphology
47:1-15.

Herring, S. W. 1993a. Formation of the vertebrate face: epigenetic
and functional influences. American Zoologist 33:472-483.

.1993b. Epigenetic and functional influences on skull
growth. /n J. Hanken and B. K. Hall. The Skull, vol. 1. University
of Chicago Press, Chicago.

Holland, P. W. H. 1992. Homeobox genes in vertebrate evolution.
BioEssays 14:267—273.

Hopson, J. A. 1979. Paleoneurology. /n C. Gans (ed.) Biology of
the Reptilia 9:39—-146.

Hoyte, D. A. N. 1966. Experimental investigations of skull mor-
phology and growth. International Review of General and Ex-
perimental Zoology 2:345—407.

. 1971. Mechanisms of growth in the cranial vault and base.
Journal of Dental Research 6(Supplement): 1447-1461.

TIMOTHY ROWE

——. 1975. A critical analysis of the growth in length of the
cranial base. Birth Defects: Original Article Series 11:255—282.

Huber, W. 1957. Analyse experimentale des facteurs topogenetiques
que regissent la fromation de lencephale de la téte chez l’em-
bryon de poulet. Archives de Anatomie Microscopique et de Mor-
phologie Experimentale 46:325—405.

Hurov, J. 1986. Soft-tissue bone interface: how do attachments of
muscles, tendons and ligaments change during growth? Journal
of Morphology 189:313—-325.

Jelinek, R. and T. Pexieder. 1968. The pressure of encephalic fluid
in chick embryos between the 2nd and 6th day of incubation.
Physiologia Bohemoslovaca 17:297—305S.

—. 1970a. Pressure of the CSF and the morphogenesis of the
CNS. I. Chick embryo. Folia Morphologica 18:102—110.

-1970b. Pressure of the CSF and the morphogenesis of the
CNS. II. Pressure necessary for normal development of brain
vesicles. Folia Morphologica 18:184—192.

Jerison, H. J. 1973. Evolution of the Brain and Intelligence. Aca-
demic Press, New York.

. 1990. Fossil brains and the evolution of the neocortex.
Pages 5-19 in B. L. Finlay, G. Innocenti and H. Scheich, editors.
The Neocortex: Ontogeny and Phylogeny. Plenum Press, New
York.

Kemp, T.S. 1979. The primitive cynodont Procynosuchus: func-
tional anatomy of the skull and relationships. Philosophical Trans-
actions of the Royal Society of London (B) 285:73—122.

— 1980. The primitive cynodont Procynosuchus: structure,
function and evolution of the postcranial skeleton. Philosophical
Transactions of the Royal Society of London (B) 285:217-258.

. 1982. Mammal-like Reptiles and the Origin of Mammals.
Academic Press, London.

———. 1983. The relationships of mammals. Zoological Journal
of the Linnean Society 77:353-384,

Kermack, D. M. and K. A. Kermack. 1984. The Evolution of Mam-
malian Characters. Croom Helm, London.

Kermack, K. A., F. L. Mussett, and H. W. Rigney. 1981. The skull
of Morganucodon. Zoological Journal of the Linnean Society
71:1-158.

Keynes, R. and A. Lumsden. 1990. Segmentation and the origin
of regional diversity in the vertebrate central nervous system.
Neuron 2:1—9.

Kielan-Jaworowska, Z. 1986, Brain evolution in Mesozoic mam-
mals. Contributions to Geology, University of Wyoming 3:21-34.

Kluge, A. G. 1988. The characterization of ontogeny. Pages 57-81
in C. J. Humphries, editor. Ontogeny and Systematics. Columbia
University Press, New York.

BRAIN HETEROCHRONY

Kraus, D. B. and B. H. Fadem. 1987. Reproduction, development
and physiology of the gray short-tailed opossum (Monodelphis
domestica). Laboratory Animal Science 37:478-482.

Laing, N. G. 1982. Abnormal development of vertebrae in paralyzed
chick embryos. Journal of Morphology 173:179-184.

Langile, R. M. and B.K. Hall. 1993. Pattern formation and the
neural crest. Pages 77-111 in J. Hanken and B. K. Hall, editors.
The Skull. Volume I. University of Chicago Press, Chicago.

Langworthy, O. R. 1928. The behavior of pouch-young correlated
with the myelinization of tracts in the nervous system. The Journal
of Comparative Neurology 46:201—248.

Larsell, O., E. McCrady Jr., and A. A. Zimmermann. 1935. Mor-
phological and functional development of the membranous la-
barynth in the opossum. The Journal of Comparative Neurology
63:95-118.

Lende, R. A. 1963a. Cerebral cortex: a sensorimotor amalgam in
the Marsupialia. Science 141:730—732.

-. 1963b. Motor representation in the cerebral cortex of the
oposum (Didelphis virginiana). The Journal of Comparative Neu-
rology 121:405—415.

. 1963c. Sensory representation in the cerebral cortex of the
oposum (Didelphis virginiana). The Journal of Comparative Neu-
rology 121:395—403.

Lloyd, D. and E. L. Rossi. 1993. Biological rhythms as organization
and information. Biological Reviews of the Cambridge Philo-
sophical Society 68:563—577,

Maddison, W. P. and D. R. Maddison. 1992. McClade: Analysis of
Phylogeny and Character Evolution (version 3.0), Sinaur and As-
sociates, Sunderland.

Maier, W. 1987. Der Processus angularis bei Monodelphis domes-
tica (Didelphidae; Marsupialia) und seine Beziehungen zum Mit-
telohr: eine ontogenetische und evolutions-morphologische Un-
tersuchung. Morphologisches Jahrbuch 133:123—-161.

Malacinski, G.M. 1984. Axis specification in amphibian eggs.
Pages 435-465 in G.M. Malacinski and S. V. Bryant, editors.
Pattern Formation. Macmillan, New York.

Marx, J. 1992. Homeobox genes go evolutionary. Science 255:399—
401.

McClain, J. A. 1939. The development of the auditory ossicles of

the opossum (Didelphys virginiana). Journal of Morphology
64:21 1-265.

McClay, D.R. and C. A. Ettensohn. 1987. Cell adhesion in mor-
phogenesis. Annual Review of Cell Biology 3:319-345.

McCrady, E., Jr. 1938. American Anatomical Memoirs, No. 16:
The Embryology of the Opossum. Wistar Institute of Anatomy

and Biology, Lancaster.

McCrady, E., Jr., E.G. Wever, and C. W. Bray. 1937. The devel-

opment of hearing in the opossum. Journal of Experimental Zo-
ology 75:503-517.

McKenna, M. C. 1975. Toward a phylogenetic classification of the
Mammalia. Pages 21-46 in W. Luckett and F. S. Szalay, editors.
Phylogeny of the Primates. Plenum Publishing Company, New
York.

Metec!f, M. E. M. and R. B. Borgens. 1994. Weak applied voltages
interiere with amphibian morphogenesis and pattern. Journal of
Experimental Zoology 268:323-338.

Metcalf, M. E.M., R. Shi, and R. B. Borgens. 1994. Endogenous
ionic currents and voltages in amphibian embryos. Journal of
Experimental Zoology 268:307—322.

Miao, D. 1990. On the origin of mammals. Pages 579-597 in H.-
Schultze and L. Treub, editors. Origins of the Higher Groups of
Tetrapods. Cornell University Press, Ithaca.

Mittenthal, J. E. 1989. Physical aspects of the organization of de-
velopment. Complex Systems (D. Stein, ed.) 10:491—S28.

Moss, M. L. 1958. Rotations of the cranial components in the grow-
ing rat and their experimental alteration. Acta Anatomica 32:65
86.

———. 1968. The functional matrix. Pages 85—98 in B. Kraus and
R. Riedel, editors. Vistas in Orthodontics. Lea and Febiger, Phila-
delphia.

Murray, P. D. F. and D. B. Drachman. 1969. The role of movement
in development of joints and related structures: the head and neck
in the chick embryo. Journal of Embryology and Experimental
Morphology 22:349-371.

Nanagas, I. C. 1925. Hydrocephalus artificially produced and its
importance in research. Journal of the Philippine Islands Medical
Association 5:251—262.

Newman, S.A. and W.D. Comper. 1990. "Generic" physical
mechanisms of morphogenesis and pattern formation. Develop-

ment 110:1—18.

Northeutt, R. G. 1984. Evolution of the vertebrate central nervous

system: patterns and processes. American Zoologist 24:701—716.
Olson, E.C. 1959. The evolution of mammalian characters. Evo-

lution 13:344-353.

Oster, G. F., J.D. Murray, and M. K. Main. 1985. A model for
chondrogenic condensations in the developing limb: the role of
extracelur matrix and cell tansactions, Journal of Embryology
and Experimental Morphology 89:93—112

Oster, G. F., N. Shubin, J.D. Murray, and P. Alberch. 1988. Evo-
lution and morphogenetic rules: the shape of the vertebrate limb
in ontogeny and phylogeny. Evolution 42:862—884.

Pacheco, M. A., W.R. Marks, G.C. Schoenwolf, and M. E. Des-
mond. 1986. Quantification of the initial phases of rapid brain
enlargement in the chick embryo. American Journal of Anatomy
175:403-411.

Patterson, B. and E. C. Olson. 1961. A triconodontid mammal from
the Triassic of Yunnan, Pages 129-191 in G. Vandebroek, editor.
International Colliquium on the Evolution of Lower and Non
Specialized Mammals. Volume 1. Koninklijke Vlaamse Aca-
demie voor Wetenschapen, Letteren en Schone Kunsten van Bel-
gie, Brussels.

Pritchard, J. J. 1972. The control or trigger mechanism induced by
mechanical forces which causes responses of mesenchymal! ~ells
in general and bone apposition and resorption in particula’’ ‘Acta
Morphologica Nerlando-Scandanavica 10:63—69.

Quiroga, J. 1980. The brain of the mammal-like reptile Prohainog-
nathus jenseni (Therapsida, Cynodontia), A correlative paleo-neu-
rological approach to the neocortex at the reptile-mammal tran-
sition. Journal fir Hirnforschung 21:299-336.

. 1984. The endocranial cast of the advanced mammal-like
reptile Therioherpeton cargnini (Therapsida-Cynodontia) from
the Middle Triassic of Brazil. Journal fiir Himmforschung 25:285—
290.

Rakic, P. 1974. Neurons in rhesus monkey visual cortex: systematic
relation between time of origin and eventual disposition.
Science 183:425—427.

——— 1988. Specification of cerebral cortical areas. Science
241:1 70-176.

Reig, O. A., J. A. W. Kirsch, and L. G. Marshall. 1987. Systematic
relationships of the living and Neocenozoic American “opossum-
like” marsupials (Suborder Didelphimorphia), with comment on
the classification of these and of the Cretaceous and Paleogene
New World and European metatherians. Pages 1—89 in M. Archer,
editor. Possums and Opossums: Studies in Evolution. Surrey
Beatty and Sons, Sydney.

Reiner, A. 1991. A comparison of neurotransmitter-specific and
neuropeptide-specific neuronal cell types present in the dorsal
cortex in turtles with those present in the isocortex of mammals:
implications for the evolution of the isocortex. Brain Behavior
and Evolution 38:53-91.

Riese, W. 1945. Structure and function of the brain of the opossum
(Didelphis virginiana) at the time of birth. Journal of Mammalogy
26:148-153

Romer, A. S., and L. W. Price. 1940. Review of the Pelycosauria.
Geological Society of America Special Papers 28:1—S38.

Rowe. T. 1986. Homology and evolution of the deep dorsal thigh
musculature in birds and other Reptilia. Journal of Morphology
189:327-346

1988. Definition, diagnosis, and origin of Mammalia. Jour-
nal of Vertebrate Paleontology 8:241—264.

1993. Phylogenetic systematics and the early history of
mammals. Pages 129-145 in F. S. Szalay, M. J. Novacek and M.
C. McKenna, editors. Mammalian Phylogeny. Springer Verlag,
New York

In press. Co-evolution of the mammalian middle ear and
neocortex. Science

TIMOTHY ROWE

_ In press. Evolutionary Biology of the Opossum. University
of Texas Press, Austin (CD-ROM for MS Windows platform).

Rowe, T. and J. Gauthier. 1992. Ancestry, paleontology, and the
definition of the name Mammalia. Systematic Biology 41:372—
378.

Rowe, T., W. Carlson, and W. Bottorff. 1993. Thrinaxodon: digital
atlas of the skull. University of Texas Press, Austin (CD-ROM
for MS-DOS platform).

———. 1995. Thrinaxodon: digital atlas of the skull. University of
Texas Press, Austin (2nd edition, CD-ROM for MS Win-
dows/Macintosh platforms).

Rubenstein, J. L. R., S. Martinez, K. Shiamura, and L. Puelles. 1994.
The embryonic vertebrate forebrain: the prosomeric model. Sci-
ence 266:578-580.

Simpson, G. G. 1927. Mesozoic Mammalia. IX. The brain of Ju-
rassic mammals. American Journal of Science 214:259-268.

——— 1959. Mesozoic mammals and the polyphyletic origin of
mammals. Evolution 13:405—414.

Smith, K. K. 1992. The evolution of the mammalian pharynx. Zoo-
logical Journal of the Linnean Society 104:313-349.

Spyropoulos, M. N. 1978. The morphogenetic relationship of the
temporal muscle to the conronoid process in human embryos and
fetuses. American Journal of Anatomy 150:395—410.

Steinberg, M.S. 1978. Specific cell ligands and the differential ad-
hesion hypothesis: how do they fit together. Pages 97-130 im D.
R. Garrod, editor. Specificity of Embryological Interactions.
Chapman and Hall, London,

Storey, E. and S. A. Feik. 1986. Remodeling of bone and bones:
effects of altered mechanical stress on the regeneration of trans-
planted bones. Anatomical Record 215:153—166.

Tan, S.-S. and S. Breen. 1993. Radial mosaicism and tangential
cell dispersion both contribute to mouse neocortical development.
Nature 362:638—640

Thatch, W. T., H. Goodkin, and J. G. Keating. 1992. The cerebellum
and the adaptive coordination of movement. Annual Review of
Neuroscience 15:403-442.

Thompson, D°A. W. 1942. On Growth and Form, 2nd Edition, Cam-
bridge University Press, Cambridge.

Toeplitz, C. 1920. Bau und Entwicklung des Knorpelschadels von
Didelphyvs marsupialis. Zoologica (Stuttgart) 27:1—84.

Ulinski, P. S. 1971. External morphology of pouch young opossum
brains: a profile of opossum neurogenesis. The Journal of Com-
parative Neurology 142:33—S8.

—— 1986. Neurobiology of the theraspid-mammal transition.
Pages 149-171 in N. Hotton If, P. D. MacLean, J. J. Roth and
E.C. Roth, editors. The Ecology and Biology of Mammal-like
Reptiles. Smithsonian Institution Press, Washington.

BRAIN HETEROCHRONY

Van Limborgh, J. 1972. The role of genetic and local environmental
factors in the control of postnatal craniofacial morphogenesis.
Acta Morphologica Nerlando-Scandinavica 10:37-47.

Wake, D. B., P. M. Mabee, J. Hanken, and G. Wagner. 1993. De-
velopment and evolution — the emergence of a new field. Pages
582-588 in E. C. Dudley, editor. The Unity of Evolutionary Bi-
ology: Proceedings of the Fourth International Congress of Sys-
tematic and Evolutionary Biology. Dioscorides Press, Portland.

Walsh, C. and C. L. Cepko. 1992. Widespread dispersion of neuro-
nal clones across functional regions of the cerebral cortex. Science
255:434 440.

Weed, L. H. 1920. The experimental production of an internal hy-
drocephalus. Contributions to Embryology, Carnegie Institution
of Washington 9:425—466.

Wible, J. R. 1991. Origin of Mammalia: the craniodental evidence
reexamined. Journal of Vertebrate Paleontology 11:1—28.

Wible, J. R. and J. A. Hopson. 1993. Basicranial evidence for early
mammal phylogeny. Pages 45—74 in F. S. Szalay, M. J. Novacek
and M. C. McKenna, editors. Mammalian Phylogeny. Springer
Verlag, New York.

Wilkenson, D. G., S. Bhatt, M. Cook, E. Boncinelli, and R. Krum-
laur. 1989. Segmental expression of Hox-2 homeobox-containing
genes in the developing mouse hindbrain. Nature 341:405—409.

Wong, M. and D.R. Carter. 1990. A theoretical model of endo-
chondral ossification and bone architecture construction in long
bone ontogeny. Anatomy and Embryology 181:523—532.

Young, R. W. 1959. The influence of cranial contents on postnatal
growth of the skull in the rat. American Journal of Anatomy
105:383-415.

Zeller, U. 1993. Ontogenetic evidence for cranial homologies in
monotremes and therians, with special reference to Ornithorhyn-
chus. Pages 95-107 in F.S. Szalay, M. J. Novacek and M. C.
McKenna, editors. Mammalian Phylogeny. Springer Verlag, New
York.

List of Abbreviations

als - alisphenoid

ang - angular

ar - articular

bas - basisphenoid

boc - basioccipital

co - cochlea

ec - endocranial cavity

fv. - fenestra vestibuli

h - standardized height of endocranial cavity
ipa - interparietal

j - jugal

Id - lambdoidal crest

mas - mastoid region of petrosal
mx - rock matrix surrounding parts of 7hrinaxodon skull
mx/ec - rock matrix in endocranial cavity
¢ - occipital condyle

opth - opisthotic

pa - parietal

pet - petrosal

pin - pineal foramen

ppr - paroccipital process

pr - promontorium of petrosal
pra - prearticular

q - quadrate

qj - quadratojugal

sag - sagittal crest

sang - surangular

sq - squamosal

st - stapes

tr - tympanic recess

tym - ectotympanic

EVOLUTIONARY DEVELOPMENTAL BIOLOGY —
PROSPECTS FOR AN EVOLUTIONARY SYNTHESIS
AT THE DEVELOPMENTAL LEVEL

David B. Wake

Museum of Vertebrate Zoology and
Department of Integrative Biology
University of California
Berkeley, California 94720

The relationship of development to evolution has engaged
the interest of biologists for well over a century, but the
subject has been something of a “fringe” issue for both fields.
The famous “three-fold parallelism” between comparative
anatomy, embryology, and paleontology predates darwinism,
and the issue seems to re-emerge with each scientific gen-
eration. Although de Beer (1930) produced a thoughtful book
on evolution and development at the beginning of the period
of the evolutionary synthesis, it was not seen as being pro-
gressive by later workers. Rather, he is credited with having
led the rejection of a causal relation between ontogeny and
phylogeny, and the neoLamarckian ideas prevalent in the
1930s, evident in the work of MacBride and his students
(Churchill, 1980). Embryology has been dismissed as being
of only passing interest in the development of the evolution-
ary synthesis (Mayr and Provine, 1980); Hamburger (1980)
refers to the absence of embryology in the synthesis as “the
missing chapter,” and suggests that only Schmalhausen,
whose work did not become available to western scientists
until 1949, could have provided it. Dobzhansky, too, felt that
Schmalhausen’s work had the potential of completing the
evolutionary synthesis (Gilbert, 1994), but such was not to
be the case, for Schmalhausen was silenced by the era of
Lysenkoism in the Soviet Union and his work has not re-
ceived the recognition it deserves until recently (Wake,
1986). While Huxley and de Beer, and later, Waddington,
were concerned with the development-evolution relationship
and made important contributions, the promise of a synthesis
has yet to be attained. My thesis in this essay is that technical
and conceptual advances have at last made such a synthesis
an achievable goal.

Despite much excellent work by earlier workers, I believe
that it was the publication of Gould’s (1977) Ontogeny and
Phylogeny, which appeared during a period of renewed in-
terest in macroevolutionary phenomena, that stimulated de-
velopment of an on-going research program into the rela-
tionship of development to evolution. His was an historical
summary of the idea of recapitulation and of attempts to use
heterochrony as a general explanation for departures from
recapitulation. In it Gould presented his own effort to model
heterochrony (the “clock-face model”) and analyze it in evo-
lutionary, ecological and paleontological perspective. In short
order, his model was reformulated in a manner that made
heterochrony accessible as an area of research (Alberch et

New Perspectives on the History of Life
Editors, M. T. Ghiselin and G. Pinna

al., 1979), and Alberch (1980) threw down the gauntlet in
challenging evolutionary biologists to be more conscious of
new findings in developmental biology. A Dahlem Confer-
ence soon followed (Bonner 1982), and a textbook was pub-
lished (Raff and Kaufman 1983); the next decade witnessed
the publication of many research papers, at levels of organi-
zation ranging from molecules to whole organisms. At the
end of the decade a period of consolidation occurred, with
several more conferences taking place (Miller et al., 1989;
Geobios Mem. Spec., 1989; Wake et al., 1991) and three
major textbooks appearing (John and Miklos, 1988; McKin-
ney and McNamara, 1991; Hall, 1992). The topic of devel-
opment and evolution had become a standard expectation of
conferences, and summaries of research activity dealing with
specific groups are appearing (e.g., Raff, 1992a).

There was a solid intellectual foundation, reviewed by
Gould (1977), on which to develop the new research initia-
tives. Following de Beer’s (1930) reanalysis of the biogenetic
law of Haeckel, there were several attempts to bring the on-
togeny/phylogeny relationship into evolutionary biology.
While conducting my doctoral studies on salamander osteol-
ogy (Wake, 1966) I became intrigued by the possibility that
some morphological features of systematic significance
might have a relatively simple developmental basis (a fine
example is the demonstration by Alberch and Gale [1985]
that whether frogs or salamanders have four or five toes can
be determined mainly by the number of cells and rate of
cell proliferation). Students of salamanders always have be-
fore them the specter of the axolotl, and so learn early that
to ignore the possibility of paedomorphosis is to imperil
one’s interpretations. Accordingly, many of us focused at-
tention on heterochrony, which had in the past played such
an important role in evolutionary narratives (e.g., origins of
vertebrates), Comparative analysis of ontogenies of particular
features within the context of the whole organism and its
ecology and biogeography (as in my studies of the premax-
illary bone, Wake, 1966, 1991) gave promise for under-
standing how morphology evolves.

The demonstration of the importance of hierarchical inter-
actions in ecology, development and evolution (e.g., com-
munity dynamics giving rise to selection on growth rate or
adult body size; genome size at the cell level impacting on
growth and differentiation rates; Wake, 1991) has many im-
plications. These have been studied from such diverse per-

spectives as quantitative genetics (Atchley and Hall, 1991),
genome size-cell size-histogenesis (Roth et al., 1990; Roth
et al., 1994), genes in relation to morphology (Nijhout, 1990),
and molecular and developmental genetics in relation to
ground plans (Slack, et al., 1993; Garcia-Fernandez and Hol-
land, 1994). Hierarchical analysis has been important in my
own work (Wake, 1991; Roth et al., 1994), and I believe
the approach is fruitful when exploring the interaction of
developmental and evolutionary processes that lead to phy-
logenetic patterns.

Rather than attempting to summarize an active and dy-
namic area of research that is changing quickly, I will discuss
prospects by focusing on some recent research trends. I first
identify some central themes and then present a selection of
topics that appear to have promise.

Central Themes

Researchers in the fields of evolutionary and developmen-
tal biology both deal with pattern and process, but terms
have been made to serve too many masters; pattern and proc-
ess become conflated, and muddle follows. Process as per-
ceived by evolutionary biologists starts with natural selection,
but increasingly is seen as extending to the formation of new
evolutionary units (lineage origination, establishment of spe-
cies), Whereas pattern emerges from cladogenesis, extinction,
and events related to earth history (mass extinctions, plate
tectonics, etc.). In development, process includes a wide
spectrum of phenomena, from genetic signaling and autono-
mous cell activities (e.g., assembly of cytoskeleton, mitosis)
to integration (e.g., contact inhibition, adhesion) (Wessells,
1982). As far as developmental pattern is concerned, pattern
formation, morphogenesis, and even ground plans are much
discussed in developmental biology. There is, however,
something of a cultural and sociological gap between primary
researchers in evolutionary, as contrasted with developmen-
tal, biology (Raff, 1992b). Evolutionists interested in devel-
opment, especially systematists and phylogeneticists, focus
on evolutionary patterns, while developmentalists are more
concerned with processes and mechanisms. The mechanisms
in evolutionary biology relate to changes in gene frequency
(population genetics) and to the effects of many genes on
overall morphology (quantitative genetics), while those in
developmental biology relate to the genetic basis of pattern-
ing in early development, to genetic signaling and cell-cell
interactions, etc. Accordingly, there is neither a common
theme nor a common vocabulary. Furthermore, while terms
are important to evolutionary biologists, developmental bi-
ologists are more pragmatic and are unlikely to debate what
is meant by the term “gastrulation,” for example; they look
instead for common themes in gastrulation. Evolutionary bi-
ologists, in contrast, disagree on even the most fundamental
concepts (e.g., homology, species), which dooms the search
for simple mechanisms and makes it difficult to find common
themes.

On the other hand, developmental biologists interested in
evolution often accept a rather simplistic notion of what an
evolutionary term (e.g., Bauplan) is, and as a result the ap-

DAVID B. WAKE

plication of their sophisticated laboratory work to questions
in evolution or phylogenetics appears to fall into the realm
of essentialism (e.g., Slack et al., 1993: Patel et al., 1994).
Developmental biologists seek commonality — the basis of
ground plans, for example; evolutionists celebrate variation
and diversity and are wary of generalizations beyond those
of formal mathematically based models (e.g., as in population
and quantitative genetics). For example, an area of continuing
controversy in evolutionary biology has to do with whether
there are macroevolutionary rules and principles, or if pat-
terns at high taxonomic levels are just the result of micro-
evolutionary processes being played out over time in an ever-
changing world.

Development meets evolution most directly in discussions
of transformation of body form. At one time, developmental
biologists believed strongly in genetic invention, and some
still do (Lovtrup, 1987). Currently there is an increased
awareness of hierarchical interactions and of generic, often
physically based, factors that regulate developmental proc-
esses (Newman and Comper, 1990; Newman, 1993, 1994),
and regulation is becoming a theme in the work of some
molecular geneticists as well (Carroll, 1994). Evolutionists
have been wary of evolutionary interpretations that relied on
genes of major effect (e.g., macromutations) since the early
1930s, but they also have been slow to adopt hierarchical
thinking, which is essential 1f modern developmental genetics
is to be understood in evolutionary and phylogenetic per-
spective. Times are changing, and the stage is being set for
mutual understanding. It may seem, with the ever-growing
literature and enthusiasm generated by those studying Hox
genes and related signaling systems, that we have entered a
new era of focus on genetic invention, but careful attention
to this large body of work reveals hierarchical perspectives
dealing with levels of organization, interaction, and regula-
tion that offer a means of connecting developmental and evo-
lutionary approaches (see, for example, Carroll, 1994).

There 1s a renewed interest in ground plans, bauplans, and
similar concepts, generated by new knowledge of the stability
of developmental systems and the generality of underlying
genetic mechanisms and signaling systems (Slack et al.,
1993, Patel, 1994). Such interpretations are only possible
within a phylogenetic perspective. Similarly, there is renewed
interest in what might be termed the limits and conceptuali-
zation of “sameness,” and this has revived interest in the
ancient debate over homology (Hall, 1994). This debate in-
volves central issues in modern developmental genetics
(Wake, 1994), for which phylogenetic perspectives are cru-
cial.

The relation of ontogeny and phylogeny — sorting
pattern from process

Recapitulation will not die as an evolutionary concept, and
as a grand generalization it has its ardent supporters (e.g.,
Nelson 1978). Recapitulation occurs when two taxa being
compared share all of their ontogeny (1-e., embryos proceed
through time along a definite and knowable pathway, char-
acteristic of the lineage, termed an ontogenetic trajectory),

EVOLUTIONARY DEVELOPMENTAL BIOLOGY

except for the terminal stages. In a recent re-evaluation of
the idea, Mayr (1994:231) states clearly, “The observation
that the embryo in the development of its organs goes, seem-
ingly unnecessarily, through certain embryonic stages found
also in its remote ancestors is an undeniable fact, and must
be explained in terms that are neither metaphysical nor purely
proximate, but through a conceivable evolutionary scenario.”
The explanation, Mayr argues, lies in the facts of develop-
ment — the inducing capacities of surrounding embryonic
tissues for a somatic program. The somatic program, in com-
bination with additional nuclear genes, directs the develop-
ment of organisms. The example given is the gill arch system
of amniotes, which is functionless as a respiratory system
but of critical importance in the development of the head
and neck. In his view, recapitulation is irregular, and the
explanation is that parts not being used in some manner are
quickly lost. This is a strict functionalist, adaptationist ex-
planation fully within the neodarwinian tradition.

However, other workers, while accepting the idea of com-
plicated interactions between parts during ontogeny, focus
more on phylogenetic issues. Ontogenies are perceived as
being predictable and constrained within lineages, suggesting
that they are not mosaic but highly integrated. Evolution oc-
curs mainly by the addition of novel traits to the terminal
stages of ontogenetic trajectories (which are repeated, or re-
capitulated, from ancestor to descendant, differing only with
respect to the novel added feature), was formalized by Al-
berch el al. (1979) as a phylogenetic manifestation of pera-
morphosis, defined as the morphological expression of a par-
ticular cell-level process. There is a rich terminology. Gould
(1977) and subsequently Alberch et al. (1979) and others
redefined old terms as names for “pure” processes, such as
hypermorphosis —a temporal extension of a developmental
process, and acceleration — an increase in the rate at which
a developmental process proceeds). Thus, process (such as
cell division rate) was linked to, but discrete from pattern
(the appearance of novel morphology in a descendant, rela-
tive to an ancestor). The extreme alternative is reverse re-
capitulation, a phylogenetic manifestation of paedomorpho-
sis, the morphological expression of a particular process
(among which are progenesis —a temporal truncation of de-
velopment, and neoteny —a decrease in the rate at which a
development proceeds). In these cases descendants do not
show novelty, but rather re-express as adult states conditions
found in the ontogeny of ancestors (in practice, phylogeneti-
cists assess conditions in sister taxa and appropriate out-
group taxa, for ancestors can only be inferred, not known).

“Reverse recapitulation” is a rather cumbersome term, but
it conveys the impression of derivative taxa “backing down,”
phylogenetically, an ontogenetic trajectory that 1s charac-
teristic of related taxa by failing to complete it developmen-
tally. Mayr (in litt.) objects strongly to the concept of reverse
recapitulation. It was introduced in a strictly formal sense
by Alberch et al. (1979), in order to idealize a symmetry
between evolution by terminal addition versus elimination
of terminal stages. Perhaps “incomplete recapitulation”
would be a more neutral term.

Patterns of both recapitulation and reverse recapitulation
are kinds of heterochrony, phylogenetic differences among
taxa with respect to the timing of developmental processes.
Heterochrony itself is often treated as a very general term
(e.g., Raff and Wray, 1989; McKinney and McNamara,
1991). To treat heterochrony in such a manner has confused
evolutionary terminology by conflating developmental proc-
ess with phylogenetic pattern. For example, Raff and Wray
(1989) argue that heterochrony at the level of whole organ-
isms (“global heterochrony”) is uncommon and probably not
very significant, whereas heterochrony at the level of par-
ticular developmental and genic processes is common and
important. I would argue that it is global heterochrony that
is most likely to have phylogenetic implications, and that to
describe subtle shifts in timing of gene action as heterochrony
is to reduce the value and utility of the term. Heterochrony
with respect to whole organisms and ontogenies has its great-
est value in phylogenetics as a relatively uncommon phe-
nomenon. | have in mind such phenomena as the overall
body form of larvae versus adults, and ontogenetic trajecto-
ries of organ systems (for example, limbs), and parts (for
example, presence or absence, and shape, of skull bones) in
relation to the whole organism (e.g., Alberch and Alberch,
1981). McKinney and McNamara (1991) argued that reca-
pitulation and reverse recapitulation are unnecessary, and that
heterochrony is a sufficient inclusive term for peramorphosis
and paedomorphosis, but I disagree. Much of their book is
an attempt to make virtually everything in evolutionary bi-
ology some kind of heterochrony, thus rendering the concept
nearly empty of intellectual content. Confusion is particularly
bad with respect to heterochronic processes versus hetero-
chronic patterns. This is where terms like recapitulation and
reverse recapitulation are useful, for they are terms relating
to phylogenetic patterns.

The word “heterochrony” serves too many masters. The
term needs modification and restriction relative to its overuse
by McKinney and McNamara. A start would be to expand
our process terminology, first by distinguishing process het-
erochrony from heterotopy (differences among taxa in rela-
tive position within the embryo where developmental proc-
esses. proceed, eg., the differing results = of
ectodermal-mesenchymal interactions involving neural crest
cells in vertebrates, see Zelditch and Fink, 1995). There is
value in developing evolutionary perspectives on heteroplasy
(e.g., differing rates of cell proliferation leading to patterns
of intertaxon allometry) as well. Workers should be careful
to differentiate between developmental mechanics and proc-
esses, and phylogenetic pattern.

Reverse recapitulation is manifest in a number of taxa of
perennibranchiate but sexually mature salamanders (Siren
Ambystoma mexicanum, Necturus, Typhlomolge give exam-
ples of independent derivation of the phenomenon in four
distantly related families). These are dramatically clear cases,
in which derived taxa resemble in their overall morphology
the non-terminal ontogenetic states of ancestral or out-group
taxa. The term “perennibranchiates” refers to the fact that
these taxa retain larval-like gills and remain aquatic through-
out life. The processes responsible for the pattern have pro-

100

duced ecomorphologies (phenotypes tightly connected to par-
ticular ecologies — in the present case, aquatic versus ter-
restrial habitats) that are profoundly distinct from those of
immediate ancestors (Reilly, 1994). The genetic and devel-
opmental mechanisms underlying this evolutionary transition
from terrestrial to aquatic adults are relatively simple, for
they are readily attained. Within Ambystoma (tiger salaman-
ders, axolotl and mole salamanders) close genealogical rela-
tives may remain in the larval state or metamorphose into
terrestrial adults, and may even show polymorphism within
a species or population (Shaffer, 1993; Collins et al., 1993).
In the case of perennibranchiation, the global effects of the
phenomenon are superficially evident in morphology and
profoundly evident in ecology, but metamorphosis in sala-
manders is not a dramatic event, and perennibranchiate and
fully metamorphosed animals differ little with respect to most
organ systems. Reilly (1994) argued that we should examine
the genetic, developmental, morphological, ecological and
phylogenetic aspects of heterochronic phenomena separately
so that we can differentiate between phenotypic plasticity
(phenotypic differences caused by environmental rather than
genetic differences) and genetic fixation (at the level of spe-
cies).

Perennibranchiation is dramatic in its ecological implica-
tions, but its more long-term, evolutionary and phylogenetic
implications are less evident. In salamanders it leads not to
radiative evolution, but seems to be a dead end. There are
only two genera and a handful of species in each of the
exclusively perennibranchiate families Proteidae and Siren-
idae. In contrast, organismal-wide paedomorphosis in direct
developing salamanders of the family Plethodontidae has had
far more profound implications, judging from the combina-
tion of morphological and taxonomic diversification. Al-
though the morphological expression of the underlying de-
velopmental processes 1s superficially less evident than in
the perennibranchiate taxa, the morphological diversity en-
compasses a far greater array of morphological combinations
and associations, including substantial novelty (Wake 1966,
1991). Furthermore, although direct development is found
only in the Plethodontidae, itself only one of the ten families
of salamanders, the direct-developing taxa constitute about
two-thirds of the living species of salamanders (Wake 1966,
1987).

| first encountered paedomorphosis in direct-developing
taxa when I found taxonomic characters (mainly bones and
their parts) that varied among taxa with respect to the time
of their appearance (Wake 1966). Those characters that ap-
pear very late in ontogeny, or that are found only in the
oldest and largest members of a population or species (see
also Smirnov 1994), suggest that the taxon displays organ-
ismal-wide paedomorphosis, and so the entire morphology
and morphological ontogeny must be evaluated within this
mental framework,

Elsewhere (Wake 1989) I presented an example from sala-
manders of the genus Batrachoseps (Slender Salamanders,
family Plethodontidae) of a pattern of intertaxon hetero-
chrony. The species of this western North American lineage
all develop directly from eggs laid on land, and there is no

DAVID B. WAKE

larval stage. One infers that the morphology is paedomorphic,
in relation to out-group taxa within the plethodontid tribe
Bolitoglossini, because all species (currently there are 8 spe-
cies recognized, but several new species are currently being
described) display adult morphologies (such as a very large
cranial fontanelle, and only four toes) that represent embry-
onic or juvenile stages of out-groups and inferred ancestors.
This is not very controversial, for the logic and data are
relatively straight-forward. However, species vary with re-
spect to other traits (all late-appearing features, such as proc-
esses of bones, secondary separation of bones, and presence
or absence of bones). It is unclear whether there is a pae-
domorphocline, a sequence of derived taxa each more pae-
domorphie than the last (McNamara, 1986), with slender spe-
cies being more derived and = paedomorphic, or a
peramorphocline, a sequence of taxa showing progressively
more peramorphic characters. If the latter situation holds,
relatively robust, more fully developed (with respect to the
traits listed above) species are more derived. Having been
derived from a paedomorphic lineage (that is, one that has
shown reversed evolution), they now would be showing a
second reversal in having morphologies that resemble the
situation before the paedomorphic phase associated with the
establishment of the lineage. Other more complex hypotheses
are only slightly less parsimonious. Recent work in my lab
shows that different stages of paedomorphosis are displayed
within each of two major clades of Batrachoseps, so at least
the notion of a simple paedomorphocline can be rejected. It
is much more difficult to reject the hypothesis that there has
been a phylogenetic reversal within the clade that includes
the robust species. The very existence of heterochrony im-
ples character instability and suggests that characters are
more labile than we generally assume that they are in phy-
logenetic analyses. While robust phylogenetic hypotheses are
essential for correct interpretation, in cases such as | have
described, with organismal-wide paedomorphosis, nearly
every character becomes suspect, and there is general inse-
curity in relation to “which end is up” (1.e., character polar-
ity). To give one example from Batrachoseps, it 1s equally
parsimonious from out-group analysis as to whether presence
or absence of a prefrontal bone is ancestral. All but one of
the described species lacks the bone, but a second species
gains a tiny speck of bone in the correct position very late
in life, following achievement of sexual maturity. Are these
two species displaying an ancestral trait, or have they un-
dergone peramorphic evolution within the framework of gen-
eral paedomorphosis and restored a trait absent from their
common ancestor? | know of no way to solve this problem
by ontogenetic and phylogenetic analysis of living taxa; this
is an instance in which a better fossil record could be deci-
sive.

The reason that Hennig’s (1966) cladistic phylogenetic
procedures have been so successful among morphologists 1s
that morphological characters often persist for long periods
of time, through numerous branching events. Differential
character persistence is universal. Some systematic charac-
ters are labile phylogenetically; others, usually termed con-
servative, show greater persistence. It is easy to find exam-

EVOLUTIONARY DEVELOPMENTAL BIOLOGY

ples of high character persistence: spiral cleavage in several
major and minor taxa, the notochord and brain stem of cra-
niates, and the tripartite body plan of insects. I believe the
terminology is appropriate at all levels of the organismal
hierarchy. Thus, the Hox gene cluster is highly persistent. |
distinguish character persistence from sfasis, persistence of
the full morphology through numerous branching events
(Wake et al., 1983).

Ontogenetic trajectories display high persistence. This phe-
nomenon has been indirectly recognized by some previous
workers in a curious way — they have proposed that paedo-
morphosis is a kind of escape from specialization for lineages
(e.g., de Beer, 1930). Extreme paedomorphosis in miniatur-
ized members of various phyla has led to the loss of traits
that are considered parts of bauplans and thus deeply em-
bedded in ontogenetic trajectories (e.g., coeloms; reviewed
by Hanken and Wake, 1993).

Character persistence occurs to varying degrees across taxa
of any rank. Among salamanders, all plethodontids are lun-
gless, but only some salamandrids are. Most salamanders are
five-toed, but three genera of plethodontids have inde-
pendently become four-toed, and all species of Batrachoseps
are four-toed. There is a well-justified, general assumption
that five is the number of toes that became fixed early in
tetrapod phylogeny, and this character has had great persist-
ence, but with noteworthy exceptions that themselves have
shown persistence at another level (e.g., the two toes of ar-
tiodactyls, limblessness in various saurian clades and in
caecilians). But cladistic approaches are invalid when char-
acter persistence is low (the “flip side” of character persist-
ence is homoplasy). For example, when dealing with mito-
chondrial DNA sequences it is risky to attempt to identify
synapomorphic and symplesiomorphic substitutions in third
positions. It is easier if one is working with a coding se-
quence, such as cytochrome b, which can be analyzed cladis-
tically at the level of its encoding for amino acids, because
amino acids show much greater persistence than bases. Much
of the on-going argument over how to analyze adaptation 1s
instead an argument over character persistence. Some work-
ers (e.g., Baum and Larson, 1991) adopt an implicit premise
that adaptive traits have high character persistence, while
others (Reeve and Sherman, 1993; Frumhoff and Reeve,
1994) believe that such traits have low persistence. Organ-
ismal-wide (global) paedomorphosis is, in a cladistic sense,
a reduction in persistence in many (terminal) characters at
once, and when it is later followed by peramorphosis, con-
fusing degrees of homoplasy are encountered. In such a situ-
ation cladistic analysis can be difficult because many equally
parsimonious arrangements of taxa are possible, and the most
parsimonious may be incorrect because of false information
from many traits. This is a pattern encountered within the
large bolitoglossine clade of plethodontid salamanders, where
early in the history of the clade there apparently was general
(high persistence) paedomorphosis and this was followed by
peramorphic changes of low persistence (Wake, 1966, 1991;
Wake and Elias, 1983). A way out of this dilemma for cladis-
tic analysis is to be able to recognize organismal-wide pae-
domorphosis at a high taxonomic level, and then code indi-

10]

vidual traits accordingly.

Failure to recognize organismal-wide paedomorphosis can
have profound consequences. For generations the brains of
salamanders have been accepted as simple, generalized and
primitive with respect to their organization and degree of
histogenesis. A recent phylogenetic analysis (as recom-
mended above) has shown that, with a high degree of prob-
ability, the brains not only of salamanders but also of
caecilians and frogs are secondarily simplified, and the nerv-
ous system is only a part of an organismal-wide paedomor-
phic syndrome (Roth et al. 1993). In the most extreme cases,
which are highly derived phylogenetically, the brains take
on an appearance similar to those of early embryos of out-
groups. This is a deceptively simple interpretation, however,
because the secondary simplification is founded on derived
patterns of connectivity and organization, and represents a
mix of embryonic (1.e., paedomorphic) and derived traits,
and in some cases novel characters, not represented in out-
group taxa. Somewhat surprisingly, the most simplified
brains are not those of perennibranchiates such as Ambystoma
mexicanum, but those of direct-developing species with com-
plex behaviors such as members of the genera Batrachoseps
and Hydromantes, and cladistically basal perennibranchiates
such as Necturus (Roth et al., 1994). The mixing of cladis-
tically derived, reversed (from paedomorphosis) traits with
persistent traits can produce substantial morphological nov-
elty. This outcome, termed ontogenetic repatterning (Wake
and Roth, 1989), can affect many seemingly unconnected
traits at once, with profound implications (as in the organi-
zation of the neural control of feeding and brainstem organi-
zation in salamanders, Wake, 1993).

The relation of ontogeny to phylogeny is no longer studied
with the goal of finding phylogenetically ancestral conditions
of whole organisms or even of traits, but with the awareness
that it is the entire ontogeny of organisms that 1s subject to
evolutionary change. A character cannot be separated from
its ontogeny (de Queiroz, 1985), nor can a character be fully
separated or isolated from its organismal milieu. Nonetheless,
it is the general stability and conservatism of characters and
organisms during their ontogeny and phylogeny that encour-
ages us to believe that there are lessons to be learned for
phylogeny from ontogeny.

Does ontogeny recapitulate phylogeny? Sometimes, in a
limited way, more in some taxa, less in others, and probably
never in the extreme form envisaged by Haeckel. The degree
to which it does hold is determined only with difficulty, and
so the value of recapitulation as a general guide is very low,
as has long been recognized (e.g., de Beer, 1930). Nonethe-
less, it 1s surprising to find how readily biologists in many
fields leap to the conclusion that a trait that appears early
in ontogeny and then transforms is likely to display the an-
cestral condition at its first appearance. Traditions die slowly.
Those who would choose to use recapitulation as a premise
in their work would be well advised to study such detailed
analyses as that of Mabee (1993; see also Mabee, 1989),
who showed that for centrarchid fishes only 52% of a large
set of characters evolved by terminal addition. Hence use of
an ontogenetic criterion for determining character state po-

102

larity has low value. The problem of “which way is up” in
evolution and phylogeny will not soon disappear!

Hierarchical perspectives on development and
evolution.

Development is a hierarchical phenomenon in which a
complex of genetic signals, physiological signals, cell-cell
interactions, generic physical factors, and self-organizational
properties interact to produce an ontogeny. During the past
decade there has been enormous progress in our under-
standing of the nature of hierarchical interactions during the
development of the vertebrate head, as exemplified by the
paper in this volume by Holland (p. 63). | will briefly con-
sider this issue not from the perspective of the genetic sig-
nalling that appears to be so important in head development,
but from the perspective of an evolutionist and phylogeneti-
cist, trying to understand how heads have evolved.

That there is a relation of lox genes to neuromeres seems
indisputable, and many labs are actively engaged in research
to pinpoint the specific mechanisms and interactions that re-
late genes to morphology. The most impressive evidence that
the genes are specifically related to neuromere formation
comes from the concordance of neuromere order, the ar-
rangement of genes within gene families on chromosomes,
and the general (there are some specific exceptions) sequence
of gene expression. At points as yet undetermined, but ap-
parently within the craniate lineage (a critical need is for
more work on basal fish lineages), there has been extensive
paralogous duplication, resulting in four gene families all
showing the same general ordering, but having different spe-
cific patterns of expression (Garcia-Fernandez and Holland,
1994). This area of research is one in which the zeal to find
a common developmental genetic ground plan for the ver-
tebrate head has proceeded with only the most general kind
of comparative structure, and with little attention to variation.
That is beginning to change, as can be seen from the recent
paper by Gilland and Baker (1993), who show that within
the general pattern of conservation there is also variation in
neuromere-gene relationship. The species differences found
indicate the likelihood of shifting relations between genet-
cally determined rhombomere identity and cranial nerves.
Although the generality is impressive at this early stage of
comparative developmental genetics, the issues of homology
and conservatism versus change are likely to loom large in
the future. The need for a detailed phylogenetic analysis of
the impressive new ontogenetic data, from a broader com-
parative base than now exists, 1s critical (Meyer, 1996, pre-
sents a detailed exposition of this point).

There long has been controversy concerning the pattern of
segmentation of the vertebrate head, and whereas once the
issue Was one of how many segments were incorporated into
the head (e.g., de Beer, 1937), and later the role of novel
vertebrate developmental interactions (e.g., those involving
the neural crest; Gans and Northcutt, 1982; Northcutt and
Gans, 1982), now the issues being raised relate to differences
in numbers and identity of rhombomeres and somitomeres,
to the nature of differences in head versus body development

DAVID B. WAKE

and organization, and to differences in genome and cell size.

Whereas Gans and Northcutt argued for a new perspective
on the organization of the head and focused on novel features
associated with interactions of neural crest cells, recent work
has gone even further and has focused on major differences
in head-body origination. Holland (1996:63—70) and Fernan-
dez-Garcia and Holland (1994) have shown that it may be
more appropriate to view amphioxus as being mainly head
rather than mainly body, in terms of the pattern of gene
product distribution during development. Fritzsch and North-
cutt (1993) proposed that cranial and spinal nerves of ver-
tebrates may not be homologues, and argued that the old
view that ocular motor nerves were homologues of spinal
motor nerves and of the so-called somatic motor component
of ventral roots of brainstem mixed nerves can no longer be
strongly defended. Northcutt (1993) refutes the influential
model of Goodrich (1930), based on modification of trunk-
like segments in head origins, and argues against a close
relationship between nerves and mesodermal derivatives. He
envisions as many as four separate cranial-caudal series of
special nerves having arisen in the heads of basal vertebrates.
These were derived independently with relationship to each
of the iterative developmental tissues of the head (neural
crest, neuromeres, placodes, somitomeres), each ina different
manner. Gilland and Baker (1993) have gone further, making
comparisons between the cranial region of vertebrate em-
bryos and the primary gastrula of amphioxus. In their view
the vertebrate head is primary, the homologue of nearly the
entire gastrula of ancestral chordates, and craniogenesis dur-
ing gastrulation is the proper structural starting point for ex-
amining the critical roles of brain segmentation and of the
evolution of functional roles for the neural crest in craniates.
These new perspectives have turned old ideas around, and
now we have the image of an ancestral vertebrate head that
had to invent a body!

One would think, given these new perspectives, that there
would be more attention to the region of the craniovertebral
joint, but that is generally not the case. As I have pointed
out elsewhere (Wake, 1993), the vagus nerve (X) is of special
significance because most of the function of this nerve is
associated with the body, although it 1s derived from one to
several rhombomeres (the number is unclear and probably
varies among taxa) and gives every indication of being a
serial homologue of nerves V, VI] and IX. Conversely, the
motor nucleus of the hypoglossal nerve (XII) lies in the spinal
cord of frogs and salamanders, and exits through vertebrae,
although it serves head muscles (the origin of these muscles
from dorsal somitic muscle that migrates ventrally was of
key importance in Goodrich’s model). The motor nucleus of
the spinal accessory nerve (XI), only recently mapped in
amphibians (Roth et al., 1984; Wake et al., 1988; Ota et al.,
1987), lies in the vertebral column, outside the brainstem,
yet the nerve migrates anteriorly to exit through the head
together with the completely separate vagus. De Beer (1937)
and other workers thought amphibians had incorporated
fewer segments from the body into the head. This issue needs
to be thoroughly reevaluated given new findings relating to
head organization.

EVOLUTIONARY DEVELOPMENTAL BIOLOGY

It may be that part of the problem with relation to am-
phibian head organization arises from the fact that there are
fewer somitomeres in amphibian heads (as in elasmobranchs)
as compared with teleosts and amniotes (Jacobson, 1993). It
is equally parsimonious, with the data at hand, to argue
either: 1) that low numbers were ancestral and high numbers
have been gained independently in teleosts and amniotes, or
2) that high numbers were ancestral and that low numbers
have been evolved independently in elasmobranchs and am-
phibians, or 3) that low numbers were ancestral and have
been retained in elasmobranchs, but were evolved inde-
pendently again in amphibians. I favor the latter hypothesis,
on the grounds that various kinds of somitic tissue are present
in very low quantities in amphibians (e.g., sclerotome; Wake,
1970; Wake and Lawson, 1973), that amphibians show sec-
ondary simplification of the entire nervous system (Roth et
al., 1993), and that modern amphibians have larger than av-
erage to enormous genome and cell sizes and they probably
arose from ancestors which had large genomes and cells.
The connection of this last point is not immediately obvious.
However, somitomeres are expansion figures in early em-
bryogenesis that accumulate cells as they expand (Jacobson,
1993), and rhombomeres, like segments, are condensations
of cells that require certain numbers of cells before they
self-organize. Furthermore, rhombomeres first are laid out
segmentally and then they subdivide. I suggest that a major
factor in amphibian development (and especially salaman-
ders) has been large cell size, which has led to the reduction
and probable loss of resegmentation (often considered to be
a universal feature of vertebral formation in tetrapods), in
the trunk, reduction through failure of units to subdivide in
the posterior head, and possibly with new combinations of
the remaining iterative cell masses in the head. Thus the
number of segments in the head of salamanders and frogs
as compared with amniotes and some fishes could be the
absence not of primary segmentation but of secondary seg-
mentation, plus some amalgamation.

There is a strong positive correlation between genome size
and cell size in vertebrates, and in large-genomed taxa there
are important implications for rate of cell division, morpho-
genesis and adult morphology (Sessions and Larson, 1987;
Roth et al. 1994). This is especially true for relatively small
organisms. Miniaturization often leads to disruption of
ground plans, and in such taxa as those constituting the ma-
rine interstital fauna, secondary (in a phylogenetic sense)
simplification of adult morphology is nearly an expectation
(reviewed by Hanken and Wake, 1993). However, it has not
been generally appreciated that there is a difference between
physical size, in which organisms are compared by mass or
linear dimensions, and biological size, in which genome and
cell size in relation to physical size, within a phylogenetic
framework (in order to determine the direction of character
state change), are the important parameters. Using such cri-
teria, the physically large lungfishes (which have the largest
genomes, and cells, among vertebrates) are biologically
small, but the physically small salamanders (and some frogs)
are in effect biological miniatures. In such organisms we
should expect, and we do find, paedomorphic morphologies

103

that are secondarily simplified, but are in fact only partially
recapitulatory. Thus, the optic tectum and the tegmentum of
relatively large genomed salamanders, frogs, and lungfishes
are apparently embryonic in histological and some aspects
of neuronal structure and organization, but they have fully
adult physiological organization and neurological connectiv-
ity, comparable to less simplified out groups. Physically
small mammals such as shrews are biologically large, in com-
parison to other small but metabolically less active mammals,
but all small mammals are biologically large in comparison
with the large-genomed, metabolically slow, developmen-
tally retarded salamanders and lungfishes, some of which
are much larger than shrews in physical dimensions. Genome
size variation is not great in amniote vertebrates, but in most
organisms it is a factor that should not be overlooked.

Increasingly in developmental biology there is an appre-
ciation of the importance of cell number at critical stages of
morphogenesis, such as in the organization of early conden-
sations. Busturia and Lawrence (1994) used genetic manipu-
lations to produce Drosophila embryos with reduced num-
bers of abdominal primordial cells; such embryos were
unable to produce morphological patterns normally seen in
development, but denticle bands were fused to those in ad-
jacent segments and some rows were missing, bristles nor-
mally present were absent, and a pigment band was reduced.
In another example, molecular-level factors involved in
skeletal morphogenesis in mice were examined with respect
to the role of the gene superfamily known as transforming
growth factor beta (Storm et al., 1994). Mutations known as
brachypodism result in marked shortening of the limb skele-
ton. Tickle (1994) observed that the number of founder cells
for each limb element might be reduced in mutants, which
translates into insufficient growth that leads to digital defects.
Alberch and Gale (1985) showed that reductions in cell di-
vision rate alone can lead to the reduction in the numbers
of digits in both frogs and salamanders.

Prospects for a merger of development and evolution

At present the fields of development and evolution are
mainly separate, but there are prospects for an integration.
I have given some examples of areas of opportunity. Two
book-length treatments have appeared (Raff and Kaufman,
1983; Hall, 1992), but the first predated much of the recent
excitement in molecular genetics and the second was written
by a scientist whose work has dealt mainly with morpho-
genesis and not with the molecular biology of development.
Those who focus on molecular developmental genetics are
mainly focused on genetic invention and novelty (genes of
large effect) (e.g., Tabin, 1993; Tabin and Laufer, 1993),
whereas developmentalists such as Hall recognize the com-
plexities and hierarchical nature of development, and have
established working relationships with quantitative geneti-
cists (Atchley and Hall, 1992), who characteristically focus
on many genes, each of minor effect. I believe that the reso-
lution of arguments concerning the merits of these contrast-
ing approaches will come from comparisons of relatively
closely related taxa, for most of the research (with a few

104

notable exceptions, e.g., Raff et al., 1991 on sea urchin mor-
phogenesis, and Nihout, 1990, on butterfly wing patterns),
has dealt with too few and too distantly related taxa. There
has been a great deal of highly sophisticated developmental
genetic work in the past several years, but mainly the focus
has been on conserved systems (highly persistent ones, using
my suggested terminology). For example the discovery of
an apparent common dorsal-ventral patterning signal in in-
sects and vertebrates has renewed speculation that there was
a reversal (note the recurrent theme of genetic invention) of
the dorsal-ventral axis in some common ancestor (Holley, et
al., 1995). However, other workers are focused more acutely
on evolutionary issues. The recent work of Carroll (summa-
rized in Carroll, 1994) on the developmental-genetic basis
for differences in arthropod body plans stimulated interest
in the possibility of analyzing the influence of developmental
regulatory mechanisms underlying morphological transitions.
Such work may point the way for new investigations into
the genetic and developmental foundations of morphological
diversity within an appropriate evolutionary and phylogenetic
framework.

Evolutionary developmental biology is a field of great
promise. Barriers that formerly separated the disciplines of
development and evolution are being broken down, and new
research questions and programs are being formulated. As
the new field develops it will be increasingly necessary to
maintain communication with the core disciplines, and this
will require that practitioners understand and appreciate the
philosophical and conceptual issues in evolution (such as
arguments over homology, and methods of phylogenetic
analysis) and the methodological and strategic issues in de-
velopment (such as the hierarchical nature of the interaction
of molecular and cytological factors in morphogenesis), so
that the empirical core of the fields can be made relevant to
both. The success of the new field will depend critically on
bringing developmental approaches to the appropriate taxo-
nomic level, such as the populational and interpopulational
levels for evolutionary analysis, and the interspecific (1.e.,
intrageneric or intrafamilial) levels for phylogenetic analysis.
There are hurdles, even barriers, to be crossed 1f a develop-
ment-evolution synthesis is to be attained. Evolutionists must
be convinced that development has something to offer them,
and vice-versa. Amundson (1994:576) has offered a pene-
trating analysis of this issue, and has made clear what the
task of students of comparative ontogenies will be “to dem-
onstrate that a knowledge of the processes of ontogenetic
development is essential for the explanation of evolutionary
phenomena.” It will not be easy, as witness the recent attack
of Reeve and Sherman (1993) on the concept of develop-
mental constraint. In questioning whether my. structuralist
(Wake 1991) explanation for why small frogs and salaman-
ders often lose one toe, but a different one in the two taxa
(five in salamanders; one in frogs, following Alberch and
Gale, 1985), Reeve and Sherman argue that the functionalist
(adaptationist) approach must always be conducted first in
order to determine if any other kind of explanation 1s nec-
essary. Amundson (1994) has made a useful comparison be-

DAVID B. WAKE

tween constraint on form and constraint on adaptation that
has relevance here. In essence, Reeve and Sherman have
made a category mistake. The argument is not about con-
straint on adaptation but about constraint on form generation;
knowledge of the developmental pathways in frogs and sala-
manders enables predictions (e.g., concerning unknown or
unstudied taxa of frogs or salamanders) to be made. This
framework leads to the recognition of constraints, not on
adaptation but on form. We who are interested in a synthesis
of development and evolution must make clear that our goals
are not to replace neodarwinism, but to expand it by focusing
on form and its causes, the central problem in development.
Similarly, we must make clear to developmentalists that
study of variation and its genetic basis, and careful phylo-
genetic analysis, central issues in evolution, have relevance
in developmental biology as well.

I have argued elsewhere (Wake, 1991; Wake and Larson,
1987: Wake and Roth, 1989) that a synthesis and integration
of three perspectives on the evolution of form are required
for a full picture of the question: how do organisms evolve?
One of these perspectives is neodarwinian functionalism —
the heart of evolutionary biology. A second 1s biological
structuralism — the rules of form generation and_ transfor-
mation, deeply embedded but not totally subservient to de-
velopmental biology. The third is history — both the contin-
gencies of history and knowledge of the genealogical
relationships of lineages. | believe that we are not far from
a time when these three approaches, appropriately integrated,
will form the heart of a modern science of evolutionary de-
velopmental biology.

Conclusion

We are on the threshold of a new venture in evolutionary
biology, the long-awaited merger of studies on the ontoge-
netic production and phylogenetic transformation of organ-
ismal form. This new field is demanding, for it requires
understanding of mechanisms of development and of
evolutionary change, and perspectives on current dynam-
ics and on history, both the history of ideas and concepts,
and the one true but only partially known history of life
on this planet. I predict that the years ahead will at last
witness a fruitful synthesis that will bring new excite-
ment to developmental evolutionary biology.

Acknowledgments

I thank the organizers of this conference for the invitation
to participate in the stimulating discussions, and R. Amund-
son, F. Bashey, M. Ghiselin, J. Hanken, N. Holland, E.
Jockusch, E. Mayr, A. Meyer, S. Minsuk, W. Olson, M. H.
Wake, and an anonymous reviewer for discussion and com-
ments on the manuscript. My research has been sponsored
by the National Science Foundation and the Gompertz Pro-
fessorship in Integrative Biology at the University of Cali-
fornia at Berkeley.

EVOLUTIONARY DEVELOPMENTAL BIOLOGY
Literature Cited

Alberch, P. 1980. Ontogenesis and morphological modification.
Am. Zool. 20:653-667.

Alberch, P. and J. Alberch. 1981. Heterochronic mechanisms of
morphological diversification and evolutionary change in the
Neotropical salamander Bolitoglossa occidentalis (Amphibia:
Plethodontidae). J. Morphol. 167:249-264.

Alberch, P. and E. Gale. 1985. A developmental analysis of an
evolutionary trend: digital reduction in amphibians. Evolution
39:8-23.

Alberch, P., S.J. Gould, G. F. Oster, and D. B. Wake. 1979. Size
and shape in ontogeny and phylogeny. Paleobiol. 5:296-317.

Amundson, R. 1994. Two concepts of constraint: adaptationism and
the challenge from developmental biology. Philosophy of Science
61:556-S78.

Atchley, W.R. and B. K. Hall. 1991. A model for development
and evolution of complex morphological structures and its ap-
plication to the mammalian mandible. Biol. Rev., Camb. 66:101—
157.

Baum, D. A. and A. Larson. 199]. Adaptation reviewed: a phylo-
genetic methodology for studying character macroevolution. Syst.
Zool. 40:1-18.

de Beer, G. 1930. Embryology and Evolution. Oxford Univ. Press,
Oxford.

1937. The development of the vertebrate skull. Oxford
Univ. Press, Oxford.

Bonner, J. T., editor. 1982. Evolution and Development. Springer-
Verlag, Berlin.

Busturia, A. and P. A. Lawrence. 1994. Regulation of cell number
in Drosophila. Nature 370:561—563.

Carroll. S. B. 1994. Developmental regulatory mechanisms in the
evolution of insect diversity. Development 1994 Supplement
217-223.

Churchill, F. B. 1980. The modern evolutionary synthesis and the
biogenetic law. Pp. 112-122 in Mayr, E. and Provine, W. B. The
Evolutionary Synthesis. Harvard Univ. Press, Cambridge.

Collins, J. K. E. Zerba, and M. J. Sredl. 1993. Shaping intraspecific
variation: development, ecology and the evolution of morphology
and life history variation in ger salamanders. Genetica 89:167
183.

de Queiroz, K. 1985. The ontogenetic method for establishing char-
acter polarity and the characters of phylogenetic systematics. Syst.
Zool. 34:280-299.

Fritzsch, B. and R. G. Northcutt. 1993. Cranial and spinal nerve
organization in amphioxus and lampreys: evidence for an ances-

tral craniate pattern. Acta Anat. 148:96-109,

Frumhoff, P. C. and H. K. Reeve. 1994. Using phylogenies to test

105

hypotheses of adaptation: a critique of some current proposals.
Evolution 48:172—180.

Gans, C. and R. G. Northcutt. 1983. Neural crest and the origin of
vertebrates: a new head. Science 220:268—274.

Garcia-Fernandez, J. and P. W. H. Holland. 1994. Archetypal or-
ganization of the amphioxus Hox gene cluster. Nature 370:563
566.

Gilland, E., and R. Baker. 1993. Conservation of neuroepithelial
and mesodermal segments in the embryonic vertebrate head. Acta
Anat. 148:1 10-123.

Gilbert, S. F. 1994. Dobzhansky, Waddington, and Schmalhausen:
embryology and the modern synthesis. Pp. 143-154 in M.B.
Adams, editor. The evolution of “Theodosius Dobzhansky.
Princeton Univ. Press, Princeton. x1 + 249 pp.

Goodrich, E. S. 1930. Studies on the Structure and Development
of Vertebrates. Macmillan, London (reprinted 1958, Dover. New
York).

Gould, S. J. 1977. Ontogeny and Phylogeny. The Belknap Press of
Harvard Univ. Press, Cambridge.

Hall, B. K. 1992. Evolutionary Developmental Biology. Chapman
and Hall, London. x11 + 275 pp.

Hall, B. K., editor. 1994. Homology: The Hierarchical Basis of
Comparative Biology. Academic Press, San Diego. xvi + 483
pp.

Hamburger, V. 1980. Embryology and the modern synthesis in evo-
lutionary theory. Pp. 96-112 i Mary, E. and Provine, W.B
editors. The Evolutionary Synthesis. Harvard Univ. Press, Cam-
bridge.

Hanken, J. and D. B. Wake. 1993. Miniaturization of body size:
organismal consequences and evolutionary significance. Ann.
Rev. Syst. Ecol. 24:501-519.

Hennig, W. 1966. Phylogenetic Systematics. Univ. Illinois Press,
Urbana.

Holland, N. D. 1996. Homology, homeobox genes, and the early
evolution of the vertebrates. Pages 63-70 in M T. Ghiselin and
G. Pinna, editors. New Perspectives on the History of Life: Essays
on Systematic Biology as Historical Narrative. California Acad-
emy of Sciences, San Francisco.

Holley, S. A., P. D. Jackson, Y. Sasai, B. Lu, E. M. De Robertis,
F. M. Hoffmann, and E. L. Ferguson. 1995. A conserved system
for dorsal-ventral patterning in insects and vertebrates involving
sog and chordin. Nature 376:249-253.

Jacobson, A. G. 1993. Somitomeres: mesodermal segments of the
head and trunk. Pp. 42—76 in J. Hanken and B. Kk. Hall, editors.
The Skull, Vol. 1, Development. Univ. Chicago Press, Chicago

John, B. and G. Miklos. 1988. The eukaryote genome in develop-
ment and evolution. Allen and Unwin, London.

Lovtrup, S. 1987. Darwinism, the refutation of a myth. Croom
Helm, London.

106

Mabee, P. M. 1989. An empirical rejection of the ontogenetic
polarity criterion. Cladistics 5:409-416.

-. 1993. Phylogenetic interpretation of ontogenetic change:
sorting out the actual and artefactual in an empirical case study
of centrarchid fishes. Zool. J. Linn. Soc., London 107:175—291.

McNamara, kK. J. 1986. A guide to nomenclature of heterochrony.
J. Paleo. 60:4—13.

McKinney, M.L. and K. J. McNamara. 1991. Heterochrony, the
evolution of ontogeny. Plenum Press, New York. xiv + 437 pp.

Mayr, E. 1994. Recapitulation reinterpreted: the somatic program.
Quart. Rev. Biol. 69:223—232.

Mayr, E. and W.B. Provine. 1980. The Evolutionary Synthesis.
Harvard University Press, Cambridge. xi + 487 pp.

Meyer, A. 1996. The evolution of body plans: Hom/Hox cluster
evolution, model systems and the importance of phylogeny. Pp.
322-340 in P. Harvey, J. Maynard and A. J. Leigh Brown, editors,
New Uses for New Phylogenies. Oxford Univ. Press, Oxford.

Miller, G. B., G. P. Wagner, and B. K. Hall. 1989. Experimental
vertebrate embryology and the study of evolution. Report of a
workshop. Forts. Zool. 35:299--303.

Nelson, G. J. 1978. Ontogeny, phylogeny, paleontology, and the
biogenetic law. Syst. Zool. 27:324-345.

Newman, S. A. 1993, Is segmentation generic? BioEssays 1§:277—
283.

. 1994. Generic physical mechanisms of tissue morphogene-
sis: a common basis for developmental and evolutionary change
J. Evol. Biol. 7:467-488.

Newman, S.A. and W.D. Comper. 1990. “Generic” physical
mechanisms of morphogenesis and pattern formation. Develop-
ment 110:1—18

Nijhout, H. F. 1990. A comprehensive model for colour pattern
formation in butterflies. Proc. Roy. Soc. London B 229:81—113.

_ 1991. The Development and Evolution of Butterfly Wing
Patterns. Smithsonian Institution Press, Washington.

Northeutt, R. G. 1993. A reassessment of Goodrich’s model of cra-
nial nerve phylogeny. Acta Anat. 148:71—80

Northeutt, R.G. and C. Gans. 1983. The genesis of neural crest
and epidermal placodes: a reinterpretation of vertebrate origins.
Quart. Rev. Biol. 58:1—28.

Ota, Y., H. Takeuchi, M. Satou, and K. Ueda. 1987. Cobaltic lysine
study of the morphology and distribution of the cranial nerve
efferent neurons (motoneurons and preganghonic parasympa-
thetic neurons) and rostral spinal motoneurons in the Japanese
toad. J. Comp. Neurol. 259:400-423.

Patel, N. H. 1994. Developmental evolution: insights from studies
of insect segmentation. Science 266:581—-S89.

Patel, N. H., B. G. Condron, and K. Zinn. 1994. Pair-rule expression

DAVID B. WAKE

patterns of even-skipped are found in both short- and long-germ
beetles. Nature 367:429 434.

Raff, R. A. 1992a, Evolution of developmental decisions and mor-
phogenesis: the view from two camps. Development 1992 Sup-
plement: |2—22.

Raff, R. A. 1992b. Direct developing sea urchins and the evolu-
tionary reorganization of early development. Bioessays 14:21 1—
218.

Raff, R. A. and T. Kaufman. 1983. Embryos, Genes and Evolution.
Macmillan, New York.

Raff, R.A. and G. Wray. 1989. Heterochrony, developmental
mechanisms and evolutionary results. J. Evol. Biol. 2:409-434.

Raff, R.A., G. A. Wray, and J.J. Henry. 1991. Implications of
radical evolutionary changes in early development for concepts
of developmental constraint. Pp. 189-207 im L. Warren and H.
Koprowsk1, editors. New Perspectives on Evolution. Wiley-Liss,
New York.

Reeve, H. K. and P. W. Sherman. 1993. Adaptation and the goals
of evolutionary research. Quart. Rev. Biol. 68:1—32.

Reilly, S.M. 1994. The ecological morphology of metamorphosis:
heterochrony and the evolution of feeding mechanisms in sala-
manders. Pp. 319-338 mz P.C. Wainwright and S.M. Reilly,
editors. Evolutionary Morphology: Integrative Organismal Biol-
ogy. Univ. Chicago Press, Chicago. vii + 267 pp.

Reeve, H. K. and P. W. Sherman. 1993. Adaptation and the goals
of evolutionary research. Quart. Rev. Biology 68:1—32.

Roth, G. and D. B. Wake. 1985. Trends in the functional morphol-
ogy and sensorimotor control of feeding behavior in salamanders:
an example of the role of internal dynamics in evolution. Acta
Biotheor. 34:175—192.

Roth, G., J. Blanke, and D. B. Wake. 1994. Cell size predicts mor-
phological complexity in the brains of frogs and salamanders.
Proc. Natl. Acad. Sct, USA 91:4796-4800

Roth, G., D. B. Wake, M. H. Wake, and G. Rettig. 1984. Distribu-
tion of accessory and hypoglossal nerves in the hindbrain and
spinal cord of lungless salamanders, family Plethodontidae.
Neurosci. Lett. 44:53—57.

Roth, G.. K.C. Nishikawa, C. Naujoks-Manteuffel, A. Schmidt,
and D. B. Wake. 1993, Paedomorphosis and simplification in the
nervous system of salamanders. Brain Behay. Evol. 42:137-170.

Roth, G., B. Rotthuff, W. Grunwald, J. Hanken, and R. Linke. 1990.
Miniaturization in plethodontid salamanders (Caudata: Pleth-
odontidae) and its consequences for the brain and visual system.
Biol. J. Linn. Soc., London 40:165—190.

Schmalhausen, I. I. 1949, Factors of Evolution: the Theory of Sta-
bilizing Selection. Blakiston, Philadelphia.

Sessions, S. K. and A. Larson. 1987. Developmental correlates of
genome size in plethodontid salamanders and their implications
for genome evolution, Evolution 41:1239-1251,

EVOLUTIONARY DEVELOPMENTAL BIOLOGY

Shaffer, H. B. 1993. Phylogenetics of model organisms: the
laboratory axolotl, Ambystoma mexicanum. Syst. Biol.42:508—
522.

Slack, J., P. W. H. Holland, and C. F. Graham. 1993. The zootype
and the phylotypic stage. Nature 361:490-492.

Smirnov, S. V. 1994, Postmaturation skull development in Yenopus
laevis (Anura, Pipidae): late-appearing bones and their bearing
on the pipid ancestral morphology. Russian J. Herpetol. 1:21—29.

Storm, E. E., T. V. Huynh, N. G. Copeland, N. A. Jenkins, D. M.
Kingsley, and S-J, Lee. 1994. Limb alterations in brachypodism
mice due to mutations in a new member of the TGFB-superfamily.
Nature 368:639-643.

Tabin, C. 1993. Why we have (only) five fingers per hand: Hox
genes and the evolution of paired limbs. Development 116:289—
296.

Tabin, C. and E. Laufer. 1993. Hox genes and serial homology.
Nature 361:693-693.

Tickle, C. 1994. On making a skeleton. Nature 368:587—S88.
Wake, D. B. 1966. Comparative osteology and evolution of the lun-
gless salamanders, Family Plethodontidae. Mem. So. Calif. Acad.

Ser. 4:1-111.

. 1970. Aspects of vertebral evolution in the modern Am-
phibia. Forma et Functio 3:33—60.

. 1986. Foreword. Pp. v—xu in Reprinted edition of I. I.
Schmalhausen, Factors of Evolution. Univ. Chicago Press, Chi-
cago.

———. 1987. Adaptive radiation of salamanders in Middle Ameri-
can cloud forest. Ann. Missouri Bot. Garden 74:242—264

———., 1989. Phylo
bios, Mem. Spec.

enetic implications of ontogenetic data. Geo-
2:369-378.

——. 1991. Homoplasy: the result of natural selection, or evi-
dence of design limitations? Am. Nat. 138:543-567.

107

. 1993. Brainstem organization and branchiomeric nerves.
Acta Anat. 148:124—131.

. 1994, Comparative terminology. Review of B. K. Hall, edi-
tor. Homology. The Hierarchical Basis of Comparative Biology.
Science 265:268—269.

Wake, D. B. and P. Elias. 1983 New genera and a new species of
Central American salamanders, with a review of the tropical gen-
era (Amphibia, Caudata, Plethodontidae). Contrib. Sci. Mus. Nat.
Hist., Los Angeles Co. 345:1—19.

Wake, D. B. and A. Larson. 1987. Multidimensional analysis of an
evolving lineage. Science 238:42-48.

Wake, D. B. and R. Lawson. 1973. Developmental and adult mor-
phology of the vertebral column in the plethodontid salamander
Euryeea bislineata, with comments on vertebral evolution in the
Amphibia. J. Morphol. 139:251—300.

Wake, D. B. and G. Roth. 1989. The linkage between ontogeny
and phylogeny in the evolution of complex systems. Pp. 361-377
in D. B. Wake and G. Roth, editors. Complex organismal func-
tions: integration and evolution in vertebrates. John Wiley and
Sons, Chichester.

Wake, D.B., K.C. Nishikawa, U. Dicke, and G. Roth. 1988. Or-
ganization of the motor nuclei in the cervical spinal cord of sala-
manders. J. Comp. Neurol. 278:195—208.

Wake, D. B., P. Mabee, J. Hanken, and G. Wagner. 1991. Devel-
opment and evolution — the emergence of a new field. Pp. 582
588 in E.C. Dudley, editor. The Unity of Evolutionary Biology
Vol. I. Proc. Fourth Int. Cong. Syst. Evol. Biol. Dioscorides
Press, Portland.

Wessells, N. 1982. A catalogue of processes responsible for meta-
zoan morphogenesis. Pp. 115-154 in J.T. Bonner, editor. Evo-
lution and Development. Springer-Verlag. New York

Zelditch, M. L. and W. L. Fink. 1996. Heterochrony and heterotopy
stability and innovation in the evolution of form. Paleobiology
22:241-254.

WAN MUAY
3 9088 01302 6042

ISBN 0-940228-43-2

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Memoirs of the California Academy of Sciences"

See other formats