Sunteți pe pagina 1din 37

T H E L O G I C OF M O N S T E R S :

E V I D E N C E F O R I N T E R N A L C O N S T R A I N T IN D E V E L O P M E N T A N D E V O L U T I O N

by

Pere ALBERCH*

ABSTRACT Rt;SUMI~
One of the most outstanding properties of natu- Une des propri6tds les plus 6videntes de la diver-
ral diversity is its discreteness and order. Species can sit6 naturelle est sa discontinuit6 et son organisation.
be identified and classified because of this property. Les esp6ces peuvent atre identifi6es et class6es ~ cause
There are two philosophical approaches to interpret the de cette propri6t6. Philosophiquement, il existe deux
orderliness of natural systems. These two conceptual approches qui permettent d'interprdter cette organisa-
positions, wich I refer to as "externalist" and "inter- tion des syst6mes naturels. Ces deux conceptions, que
nalist", prescribe drastically distinct methodological je qualifie d'"externaliste" et d'"internaliste" impo-
approaches. Classical neo-Darwinism falls within the sent des m6thodes d'approche tr6s diff6rentes. Avec la
"externalist" tradition, with its emphasis in natural sdlection naturelle comme principal agent d'organisa-
selection as the main ordering agent in evolution, this tion de l'6volution, le n6o-Darwinisme classique se rap-
approach basically argues that the properties of the porte/~ la tradition externaliste. Cette approche admet
physical ant biotic environment determine the selective que les param6tres physiques et biotiques de l'environ-
pressures and consequently dictate which form will be nement d6terminent seuls la pression de s61ection et,
selected over others. Therefore, the discreteness and en cons6quence, ddterminent quelle forme sera sdlec-
order of natural diversity is a direct reflection of the tionn6e. Ainsi, la discontinuit6 et l'ordre de la diver-
topography of the adaptative landscape. The interna- sit6 naturelle sont le reflet direct de la topographic du
list approach attributes some of the order observed in paysage adaptatif. L'approche internaliste attribue une
nature as the result of the emergent properties of gene- part de l'ordre naturel /~ l'6mergence de propri6tds
rative rules. I explore the latter methodological intrins6ques au matdriel vivant. J'examine cette derni6re
approach introducing data from teratological systems approche/t partir de donn6es tirdes de syst6mes tdra-
to illustrate how order exists, and classification is pos- tologiques pour illustrer comment cet ordre se mani-
sible, in clearly non-functional systems. Subsequently feste, et quelle classification est possible dans des systb-
I examine the properties of pattern generating systems mes clairement non fonctionnels. Ensuite, j'examine
and use limb development and evolution as a case study les propri6t6s de systbmes g6n6rateurs et j'utilise le d6ve-
to illustrate how we can carry out an evolutionary study loppement et l'6volution des membres pour illustrer
from an internalist perspective. I also discuss the struc- comment on peut mener une 6rude 6volutive dans une
ture, organization and properties of pattern-generating perspective internaliste. Je discute aussi la structure,
systems and the role of genes in morphological evolu- l'organisation et les propri6t6s des syst6mes g6ndrateurs
tion. I argue that is not possible to construct a genetic ainsi que le r61e des g6nes dans l'6volution morpholo-
theory of morphological evolution since genes and gique. J'admets qu'il n'est pas possible de construire
development cannot be dissociated as different levels une th6orie g6ndtique de l'6volution morphologique
of interaction. I conclude with a synthetic view of mor- puisque g6nes et d6veloppement ne peuvent atre disso-
phological evolution and propose that it is as interes- cids en rant que niveaux d'interaction diff6rents. Je ter-

* Museo Nacional de Ciencias Naturales. J. Gutierrez Abascal, 2, 28006 Madrid, ESPAGNE.

Geobios, m6moire sp6cial n 12 p. 21-57, 19 fig., 2 pl. Lyon, 1989


m 22 m

PLATE 1

Plate 1 - - Illustrations of animal diversily.


(Clockwise from upper left) Paleolithic cave paintings. Roman mosaics. Morphological variation in sterns (from
a contemporary field guide). R. Savery "Landschaft mit V6geln" (early 17th century) Kunsthistorisches Museum,
Vienna.
Illustrations de la diversit~ animale.
De gauche/t droite et de bas en haut : peinture rupestre pal6olithique ; mosaiques romaines ; tableau de R. Savery
"Landschaft mit V6geln" (ddbut du 17% Kunsthistorisches Museum, Vienna) ; variation morphologique chez les
sternes.
m 23 m

ting and illuminating to focus on invariance as on diver- mine par une vision synthdtique de l'6volution morpho-
sity. Similarly, monsters are a good system to study the logique et sugg6re qu'il est tout aussi fructueux de s'in-
internal properties of generative rules. They represent tdresser/~ l'invariance qu'~ la diversit6. Les monstres
forms which lack adaptative function while preserving sont un bon mod61e pour comprendre les propridt6s
structural order. An analysis of monsters, is a study internes des systbmes gdn6rateurs. Ils correspondent/~
of "pure f o r m " in the classical Naturphilosophie sense. des formes qui n ' o n t pas de fonction adaptative, mais
There is an internal logic to the genesis and transfor- qui conservent l'ordre structural. Une analyse des
mation and in that logic we may learn about the cons- monstruosit6s est une ~tude de la " f o r m e pure" au sens
traints on the normal. classique de Naturphilosophie. I1 existe une logique
dans la gen6se et la transformation des monstruosit6s
qui nous renseigne sur les contraintes qui existent dans
les cas normaux.

KEY-WORDS : ONTOGENY, PHYLOGENY, AMPHIBIAN.

MOTS-CLI~S : ONTOGENI~SE, PHYLOGENI~SE, AMPHIBIENS.

The earliest signs of human civilization - the Paleo- possible forms, i.e. any combination of x and y values.
lithic cave paintings - already reveal the interest of Man Each individual form corresponds to a point in this
in the diversity of organic form. Ever since - f r o m the space. If we were to plot all observed phenotypes, inclu-
mosaics of Pompei to the paintings of Brueghel and ding malformations that die during embryonic or post
Savery ; f r o m Pliny's Natural History to contemporary embryonic development, we would find that the phe-
field guides, passing through the rich tradition of notypic space is not uniformly filled by points. Instead,
Medieval Bestiaries - artists and scientists alike have populations of forms are clustered and separated from
been fascinated with the enormous variety of plant and others by empty regions. These clusters of points would
animal life (pl. 1). correspond to species, polymorphic forms or even clas-
ses of teratologies (e.g. wingless mutants in chicks or
We are so bewildered by the diversity of nature homoeotic mutations in Drosophila).
that we often forget that the world could have been of
a very different shape. H u m a n imagination has been Traditionnally, biologists in general and evolutio-
a fertile source of new forms - forms that could have nists in particular have focused on patterns of variabi-
existed but were not to be. Even in the early cave pain- lity. That is, the shape and distribution of the clouds
tings we already find composite creatures, human-beast of points in phenotypic space. In this paper, I am more
hydrids resulting from human fantasy. Mythology has concerned about the empty spaces, about the morpho-
been rich in generating new morphologies, such as mer- logies that, although conceivable are not realized. These
maids, centaurs and unicorns. Certain artists, f r o m discontinuities in phenotypic space account for the
Hieronymus Bosch to such modern surrealists as J o a n order observed in nature and they are the reason why
Miro have specifically attempted to create new mor- a classification of forms is possible. Forms can only
phological universes independent of external reference be described and arranged in a hierarchical classifica-
(pl. 2). tion (order, family, genus, species...) if they are dis-
crete. If populations of forms smoothly blended one
A comparison between the forms illustrated in pla- into another through a series of intermediates, any mor-
tes 1 and 2 makes obvious one the most fundamental phological diagnosis of species would be arbitrary.
questions in evolutionary biology, one that underlies
the whole field : Why do we observe certain forms in The existence of continuous variation in m o r p h o -
nature and no others ? Why d o n ' t we find mermaids logy is a position defended by strict Darwinism, roo-
or multi-headed dragons ? ted in the traditional Western view of nature as a con-
tinuum (see reviews by Lovejoy 1936 ; Mendelsohn
The idea of limited diversity can be pictorially des- 1980) and exemplified by the following quotes :
cribed using the concept of phenotypic space (Alberch
1980). Let us assume that any f o r m can be completely " ( T h e term species is) arbitrarily given, for the
described by a set of measurements corresponding to sake of convenience, to a set of individuals closely
a point in a multidimensional space (for simplicity, I resembling each other ; it does not essentially differ
will reduce this analysis to two dimensions, x and y) from the term variety, which is given to less distinct
(fig. 1). The phenotypic space is, thus, the set of all and more fluctuating forms. The term variety, again,
m 24 m

PLATE 2
Plate 2 - - Imaginary creatures.
(Clockwise from upper left). "The Sorcerer" or "Horned God", Trois Fr~res Cave (Upper Paleolithic) ; Medieval
dragon (Frontal dels Arcangels, 13 th century, Museu d'Art de Catalunya) ; Drawings from Codex Seraphinianus
(Franco Maria Ricci editor, 1981) ; Joan Miro, lithograph from the "Homenatge a Joan Prats" series, 1971).
Crl~atures imaginaires.
De gauche ~l droite et de bas en haut : "le sorcier" ou "dieu cornu", Grotte des Trois Fr6res (Pal6olithique sup&
rieur) ; dragon m~di6val (Frontal dels Arcangels, 13e si6cle, Museum d'Art de Catalunya) ; Joan Miro, lithographie
"Homenatge a Joan Prats", 1971 ; dessins tir6s de Codex Seraphinianus (Franco Maria Ricci 6diteur, 1981).
-- 25 --

in comparison with mere individual differences, is also Empirical data, however, do not support this prin-
applied arbitrarily, for convenience sake" (Darwin ciple of continuity and suggest that species are homeos-
1859, p. 42). tatic and discontinuous entities. A fact that, in most
instances, allows for the construction of an unambi-
"The word species is simply a term of convenience,
gous taxonomy. Therefore, I will begin this essay by
including such members of a group similar to each
examining the mechanisms that have been postulated
other as are tangibly different from others, and are not
to be responsible for these gaps in phenotypic space.
known to be connected with these by intermediate
forms. Such connecting links we m a y suppose in all
cases" (Jordan 1905, p. 293).

Fig. 1 - - A phenotypic space is the set of all possible phenotypes.


This diagram assumes that the morphology of an organism
can be completely defined by the measurement of two traits,
x and y. Each point corresponds to a specific morphology.
C,I ---- D Not all possible phenotypes are found. The distribution of
/~" I , "l points in non-uniform. Clusters of points could correspond
/ ~ I * ]
6 , ~
to species (e.g. A and E), polymorphic, or closely related,
>- / ""...'? ., t-, " ;" species (e.g. C1 and C2) which are not separated by dis-
,. ".'.'-..- --~- c,2 - tinct morphological discontinuities, or classes of teratolo-
F- \
gies (e.g. D and B).
\~* ",,,~**

<~ .... ", , ' * I l l ; " * r


U n espaee ph~notypique est l'ensemble de tous les ph~uoty-
Fr" \* p .. /
pes possibles.
Ce diagramme montre que la morphologie d ' u n organism
/ . : - . . \ \ I I * . . . ':, w
. - .
\ peut ~tre compl6tement d6finie par la mesure de deux carac-
~. ::lg, 1 t6res, x et y. Chaque point correspond ~. une morphologie
.,* .1 t / ' - ;'" ~* /;
particuli6re. Tousles ph6notypes possibles ne sont pas r6a-
"-'_L \ . \---" "/
tA " .', ' ":" E lis6s. La distribution des points n'est pas uniforme. Les
ensembles de points pourraient correspondre ~t des esp6-
ces distinctes (e.g. A et E), fi des esp6ces polymorphiques
ou 6troitement apparent6es (e.g. C1 et C2) qui ne seraient
pas sdpar6es par des discontinuitdes morphologiques, ou
TRAIT X enfin ~ des types de tdratologies (e.g. B et D).

A - ORIGINS OF ORDER AND DISCONTINUITY IN N A T U R E

Historically there have been two distinct approa- through the elimination of less fitted intermediate
ches to the question of why certain forms are not found forms. In this case, and in agreement with the princi-
in nature. They reflect two qualitatively different ways ple of continuity in nature, phenotypic space is effec-
of seeing the organic world. I purposely caricature the tively continuous prior to the action of selection. I refer
two positions by emphasizing their more extreme to this approach as functionalist (after Gould 1986) or
forms. At this point, I am not interested in doing an as "externalist" in the sense that the observed order
exhaustive review of the subject. Rather, I want to high- in nature is caused by agents external to the organism.
light the methodological consequences of taking each
of these philosophical stands. The origins of this school of thought can be tra-
ced back to Cuvier. In his influential writings, Cuvier
coupled the concept that functional demands essentially
1. THE EXTERNALIST/FUNCTIONALIST APPROACH
shape morphology with the hypothesis that disconti-
This is basically the thesis defended by Darwin and nuity in the history of life is the result of externally
Jordan in the quotes cited above. It claims that nature determined catastrophic events (see Appel 1987 for a
is essentially continuous. This is a philosophical rather historical review of this issue).
than empirical position. Intermediate forms, even if not
The externalist world view is so predominant in
found today, are considered possible.
the field of evolutionary morphology that it is not
Their absence can be explained by invoking his- usually made explicit. This approach views directiona-
torical contingency (e.g. they existed in the past but lity and order as imposed by natural selection acting
have disappeared since then, or, alternatively, the on variation (noise) generated by genetic mutation.
system has yet to evolve into this particular region of Selective pressures are a function of the physical and
phenotypic space). In addition, Darwinism claims that biotic factors of the environment in which the organism
natural selection acts as the main ordering agent
- - 2 6 - -

ELECTION (direction)

Fig. 2 - - E x t e r u a l i s t s c h e m e .
Phenotypic evolution is the result of the interaction of deter-
ministic natural selection acting on variation generated by
mutation.
PHENOTYPEI EVOLUTION
S c h 6 m a externaliste.
L'evolution phenotypique est le resultat de l'interaction
entre la selection naturelle d6terministe et la variation due
aux mutations.

MUTATION (noise)

is found. For example, gills are considered adaptations niches is not only immense, it is also discontinuous...
to aquatic breathing as long as there is water in the The adaptive peaks and valleys are not interspersed at
pond. Therefore, and this is what I mean by the term random. " A d j a c e n t " adaptive peaks are arranged in
externalist, the motor of evolutionary change and diver- groups, which may be likened to mountain ranges in
sification (selection) depends on factors external to the which the separate pinnacles are divided by relatively
organism (fig. 2). shallow notches. Thus, the ecological niche occupied
by the species " l i o n " is relatively much closer to those
This scheme defines a research program which occupied by tiger, puma and leopard than to those
focuses either on problems of adaptation or on the occupied by wolf, coyote, and jackal. The feline adap-
population dynamics of the system. Research on adap- tive peaks form a group different from the group of
tation is essentially a static study of how well a parti- canine " p e a k s " . . . The hierarchic nature of the biolo-
cular design performs in a particular environment rela- gical classification reflects the ascertainable disconti-
tive to alternative designs also found in the population, nuity of adaptive niches, in other words, the disconti-
in an attempt to identify selective pressures that may nuity of ways and means by which organisms that inha-
account for its present state. These studies often make bit the world derive their livelihood from the
use of machine-organism analogies and employ con- environment."
cepts of optimization borrowed from economics or
engineering models (e.g. Dullemeijer & Barel 1977). To Dobzhansky, order and hierarchical oganiza-
tion are imposed by the discontinuities in ecological
A central metaphor in the externalist approach is space ; a perfect example of the externalist approach
the concept of the adaptive landscape (Wright 1932). to phenotypic evolution. It is unlikely that Dobzhansky
An adaptive landscape is basically a phenotypic space himself would subscribe to such a radical position.
in which a fitness value is associated with every possi- Since, although there is ample empirical evidence that
ble phenotype [fitness value of phenotype (x, y) = ecological diversity plays a key role in structuring pat-
height of landscape at (x, y)]. The topography of the terns of faunal and floral diversity [e.g. Lack (1947)
landscape is, however, externally defined. As the envi- on the Galapagos (Darwin's) finches ; Williams (1972,
ronment changes, the relative fitness of the various phe- 1983) on the West Indian Lizards of the genus A n o -
notypes is affected, a fact depicted as a change in the lis ; or the abundant work on coevolution), one can-
adaptive landscape. Some adaptive peaks disappear not ignore the role of history. After all, felines are simi-
while new ones form. Organisms evolve by climbing lar not only because they occupy similar ecological
towards the top of the nearest peak. Therefore, the niches, but also because they share a common ances-
ordered variation observed in nature is a reflection of tor. In fact, it could be argued that it is because felines
the topography of the adaptive landscape. This pers- represent variations within a conserved body plan that
pective, interpreting order in terms of the properties they are constrained to a subset of similar ecological
of the adaptive landscape, is clearly illustrated by the niches. Dobzhansky's statement, nevertheless, clearly
following quote t~rom one of the fathers of the nee- illustrates the nee-Darwinian tendency to disregard his-
Darwinian synthesis, the geneticiat Theodosius Dobz- tory in favor of adaptive explanations (Gould 1983).
hansky (1951 ; pp. 9-10) : " T h e enormous diversity of In this tradition, internal factors, often equated with
organisms may be envisaged as correlated with the lack of genetic variability for the trait, may slow down
immense variety of environments and of ecological the evolutionary process but if selective pressures are
niches which exist on earth. But the variety of ecological strong enough - and population size sufficiently large
-- 27 m

they will eventually overcome any developmen- process of adaptation. The internalist approach is dia-
tal/genetic constraint (e.g. Charlesworth et alii 1982 ; grammatically depicted in fig. 3. In it, the genetic and
Lande 1986). developmental interactions that are responsible for the
genesis of biological f o r m are illustrated as a "black
2. THE INTERNALIST (STRUCTURALIST)APPROACH b o x " . More information on the nature of specific pat-
tern generating systems can be found in O s t e r &
History carries with it a set of system specific cons- Alberch (1982), and specific examples are discussed in
traints, which in turn can bias the future evolution of Appendix I. At this point it suffices to state that f o r m
the system. I refer to this philosophical tradition as is the product of physico-chemical interactions at the
internalist. This approach, instead of focusing on the genetic and epigenetic levels.
external forces of evolution, emphasizes the role of
internal factors. Internalist theories have a long, but The internalist approach assumes, and this is a key
minority, tradition in evolutionary biology (e.g. Berg- assumption, that morphological diversity is generated
son 1907; Berg 1926; Osborn 1934; Schindewolf by perturbations in parameter values (such as rates of
1936, 1950 ; Goldschmidt 1940 ; Cuenot 1951 ; Wad- diffusion, mitotic rate, cell adhesion, etc.) while the
dington 1957 ; Cannon 1958 ; Grass6 1970 ; Saunders structure of the interactions a m o n g the components
& H o 1976; Ho & Saunders 1979; Riedl 1979; remains constant. For example, a specific enzyme will
Alberch 1980 ; Rachootin & T h o m s o n 1981 ; Webster bond to its substrate with an affinity that can be quan-
& Goodwin 1982 ; Wake & Larson 1987, have defen- titatively modulated by either a genetic mutation alte-
ded the role of internal factOrs in evolutionary proces- ring the amino acid sequence at the binding site or envi-
ses f r o m a variety of perspectives). Gould (1986) has ronmental parameters, such as temperature, affecting
recently referred to this position as "structuralist" in rate of reaction. The interaction between the specific
reference to the theories of Geoffroy St. Hilaire. Given enzyme and its substrate is assumed to remain inva-
the recent, more restricted usage of the term in anth- riant, although the binding affinity could be zero.
ropology, linguistics and, even evolutionary biology Given this assumption, even if the parameters of the
(e.g. Lauder, 1981 ; Webster & Goodwin 1982), I pre- system are randomly perturbed, by either genetic muta-
fer to use the more descriptive, and philosophically less tion, environmental variance or experimental manipu-
loaded, term of "internalist". lation during development, the system will generate a
limited and discrete subset of phenotypes. Thus, the
The roots of the internalist perspective predate realm of possible forms is a property of the internal
Darwinism and they can be found in the ideas of structure of the system.
Romantic and transcendental morphologists such as
Returning to the phenotypic space in figure 1,
Goethe, G e o f f r o y St Hilaire and Owen to name only
internalist theory argues that even in the absence of
three influential figures from three different countries.
selection naturally occuring morphologies would be
These authors, and m a n y others (see historical reviews
clumped and gaps would be prominent in the distribu-
by Russel 1916 and Appel 1987), searched for invarian-
tion of morphologies. In its purest sense, the interna-
ces and universal rules of organization in support of
list approach is not interested in the problem of adap-
their basic tenet which argues that form is primary over
tation. In fact, as I will argue below, one of the best
function and that a "unity of plan" underlies the diver-
examples of internally defined order can be found in the
sity observed in nature.
organization observed in the appearance of gross mal-
These ideas can be translated, in a modern con- formations, either experimentally induced or the pro-
text, to m e a n that the regularities and trends observed duct of a genetic mutation. The internalist approach is
in phylogeny are a reflection of a conserved set of fundamentally typological. Instead of focusing on popu-
pattern-generating rules (Oster & Alberch 1982 ; Good- lation dynamics, it is concerned about the origin and
win 1982). These internal rules of development define properties of types, the limits of form, and the nature
the realm of possible variation and place limits on the of the rules that are responsible for this ordered pattern.
Fig. 3 -- Internalist scheme.
Perturbations resulting from genetic mutation or environ-
mental impact are "filtered" through the dynamics of a
PHENOTYPES pattern-generating system. The internal structure of the
developmental system defines a finite and discrete set of
possible outcomes (phenotypes)even if the sources of per-
turbation are random.
Schema internaliste.
PERTURBATION SYSTEM Les perturbations issues de mutations g6n6tiques ou de Fin-
fluence de l'environnement sont filtr6es par la dynamique
d'un syst~me"structurant". Ainsi, la structure m~med'un
d6veloppement impose un nombre fini et discret d'abou-
tissements possibles (ph6notypes), m~me si les sources de
perturbations sont al~atoires.
-- 28 --

In this paper, i explore the implications of this I focus on the evolution of form at a strictly deve-
approach and I propose a logically consistent, alter- lopmental level. It is, however, important to recognize
native way of studying morphological evolution - a that the internalist philosophy, with its emphasis on the
method rooted in an understanding of the properties structure of interactions as a system of constraints, is
of organisms rather than of the properties of the envi- not the exclusive domain of developmental biology. For
ronment. I focus on the internal rules that control the example, Garcia-Bellido (1983, 1986), Kauffman (1983)
appearance of morphological variation, on the mecha- and Agur & Kerszberg (1987) present evolutionary
nistic basis of such rules and on the evolutionary con- models based on the properties of regulatory networks
i sequences of this internally determined order. I need of genetic interactions. Similarly, Liem (1980) and Lau-
to emphasize that this approach is not mutually exclu- der (1981) have explicitely used an internalist approach
sive to the externalist or adaptationist approach (e.g. see to study the evolutionary properties of systems of func-
Larson & Wake, 1987, for an attempt to integrate both tional interactions, and Edelman (1987) incorporates
approaches). The perspective that I defend here simply the issue of developmental constraint emerging from
focuses on different issues, poses different questions and epigenetic interactions in a general theory of brain
prescribes a qualitatively different research program. function.

B. MONSTERS AS EVIDENCE FOR I N T E R N A L C O N S T R A I N T IN D E V E L O P M E N T AND E V O L U T I O N

To explore the role of internal factors in evolu- continuity of the evolutionary process defended by
tion we need first to resolve the issue of how much of Goldschmidt and others is an issue that I will not dis-
the observed pattern is internally generated. After dis- cuss in this paper.
secting internally from externally determined order, we
can proceed to address the specific genetic and deve- 1. PROPERTIES OF TERATOLOGICALSYSTEMS
lopmental mechanisms responsible for internal cons-
traint in ontogeny and phylogeny. To study the first Teratologies are a superb document of the poten-
issue we need to carry out an analysis of patterns of tiality of a given developmental process. In spite of
variation, while to address the second we have to strong negative selection, teratologies are not only gene-
understand the mechanisms that control the genesis of rated in an organized and discrete manner but they also
form during development. exhibit generalized transformational rules. These pro-
perties are not exclusive to teratology ; rather they are
When studying natural systems it is often difficult
general properties of all developmental systems. There
to distinguish between the action of selection and the
are numerous examples illustrating parallel patterns of
effect of internal constraint. For this reason, I propose
intra-populational and inter-specific morphological
that it may be advantageous to turn to non-functional,
grossly maladapted, teratologies when studying the pro- variation (e.g. Vavilov 1922 ; Spurway 1949 ; Raikow
perties of internal factors in evolution. These major 1983 ; Alberch 1983 ; Garcia-Bellido 1983 ; Shubin &
deviations from normal development result in forms Alberch in press). These properties are, however, bet-
that are often lethal, and always significantly less well ter illustrated using data from pathological systems.
adapted than their progenitors. Therefore, one expects a. Ordered and discrete generation
monsters to be consistently eliminated by selection. This
is a useful property because if, in spite of very strong Teratologies are generated in an organized man-
negative selection, teratologies are generated in a dis- ner. Certain forms are more likely to appear than
crete and recurrent manner, this order has to be a reflec- others. I discussed this property in Alberch (1980) and
tion of the internal properties of the developmental gave several examples ranging from the variation in the
system. gross anatomy of the aortic arches in rabbits to the rates
This interest in teratology contrasts with the bad of occurrence of homoeotic transformations in Dro-
reputation that monsters have traditionally enjoyed in sophila. A similar pattern is evidenced even at the level
the Darwinian literature. From an externalist perspec- of " e r r o r s " in cytodifferentiation as observed in trans-
tive, teratologies as non-functional entities are evolu- determination experiments (fig. 4a). Imaginal disks
tionary deads ends and, as such unworthy of study. I with well defined developmental fates can be tricked
emphasize that I introduce teratologies only as model into differientiating into another type of tissue when
systems to study the patterning generated by develop- cultured for long periods of time. The transformations
mental properties. Unlike the classical "hopeful mons- observed, however, are not chaotic. For example, if a
ters" of Goldschmidt (1940), I do not contend that tera- prospective genital cell makes a mistake, it will most
tologies are variation with evolutionary potential. In likely produce a leg or a head-antenna tissue type, or
fact, I assume them to be lethal in most cases. The dis- with a lower probability a palp. It will never give rise
-- 29 --

A B

Genital

Head-
leg antenna
disc disc

~ Wing disc

l
Thorax disc

Fig. 4 - - Ordered transition in developmental states.


A. Pathways of transdetermination in imaginal discs in Drosophila (from Gilbert 1985). B. Pathways of transdetermination in human
epithelial tissues (from Slack 1985). Thick arrows denote the most frequently observed events, while thin arrows depict rare ones.
Lack of arrows between differentiated states indicates that the transformation is extremely rare.
Agencement ordonn~ des ~tats dn d~veloppement.
A. Itin~raire de transd6termination des disques imaginaux chez Drosophila (d'apr~s Gilbert 1985). B. Itin&aire de transd~termina-
tion des tissus 6pithdliaux chez l'homme (d'apr6s Slack 1985). Les flbches @aisses indiquent les transformations les plus fr6quem-
ment observ6es tandis que les fl6ches minces indiquent les plus rares. L'absence de fl6che entre deux 6tats indique que la transforma-
tion est extr~mement rare.

to a wing o r a t h o r a x . Slack (1985) has recently revie- a genus and, in the w o r k o f E. G e o f f r o y Saint Hilaire,
wed a similar set o f restricted cellular t r a n s f o r m a t i o n s a species. His son, I s i d o r e o n l y classified t e r a t o l o g i e s
in h u m a n tissues (fig. 4b). T h e general result is t h a t to the generic level in his c o m p r e h e n s i v e treatise entit-
d e v e l o p m e n t a l p a t h w a y s when p e r t u r b e d b y genetic led " H i s t o i r e Gdn6rale et Particuli6re des A n o m a l i e s
m u t a t i o n o r e x p e r i m e n t a l m a n i p u l a t i o n can u n d e r g o chez l ' H o m m e et les A n i m a u x . . . " (1836), which is
o n l y a l i m i t e d set o f t r a n s f o r m a t i o n s a n d certain deve- widely r e g a r d e d as the w o r k which set up the f o u n d a -
l o p m e n t a l switches are m o r e likely to occur t h a n others. tions o f m o d e r n scientific research in t e r a t o l o g y . T h e
This p r o p e r t y was e m p h a s i z e d a n d a l l e g o r i c a l l y depic- classification p r o p o s e d b y I. G e o f f r o y St. H i l a i r e was
ted in the " e p i g e n e t i c l a n s c a p e s " o f W a d d i n g t o n n o t s u p p o s e d to reflect lineage r e l a t i o n s h i p s , i n s t e a d
(1957). it w o u l d be m o r e a n a l o g o u s to a p e r i o d i c table o f che-
mical elements or a c l a s s i f i c a t i o n o f crystal types.
I n t r i n s i c o r d e r in the g e n e r a t i o n o f m o n s t r o s i t i e s
was discussed in the p i o n n e e r i n g w o r k s o f the fathers
o f m o d e r n t e r a t o l o g y , Etienne a n d Isidore G e o f f r o y On the absence of triple monsters
Saint H i l a i r e in the early a n d m i d n i n e t e e n t h c e n t u r y .
T h r o u g h an extensive survey, these authors noticed that A n a l y s i s o f the diversity f o u n d in t e r a t o l o g i c a l
t e r a t o l o g i c a l f o r m s were n o t o n l y discrete a n d recur- f o r m s reveals the fact t h a t n o t all possible t e r a t o l o g i e s
rent, a fact t h a t allowed t h e m to be d e s c r i b e d a n d dia- are g e n e r a t e d . F o r e x a m p l e , I. G e o f f r o y Saint H i l a i r e
gnosed, b u t they also c o u l d be classified in a h i e r a r - (1836, vol. 2, chap. 12) p o i n t s out t h a t two h e a d e d
chical m a n n e r a n a l o g o u s to the L i n n e a n system o f clas- m o n s t e r s (i.e. b i f u r c a t i o n o f the a n t e r i o r region o f the
sification f o r n a t u r a l species ( e . g . I . G e o f f r o y Saint p r i m a r y axis o f the b o d y ) are relatively c o m m o n in all
H i l a i r e 1936 ; fig. 5). E v e r y m o n s t r o s i t y c o u l d be assi- g r o u p s o f vertebrates f r o m fish to h u m a n s (see below).
gned to a class and within its class to an order, a family, C o n s e r v e l y , t h r e e - h e a d e d m o r p h o l o g i e s are e x t r e m e l y
-- 30 --

~ co

-;4
or]

P--4


~p
c/3 0
. . . . ~ ~

o ~ 7

~ : ~ c ~
o

Z~

0
~ ~-~ ~ ~ . ~ ~'~ ~--~_~ . . ~ .
o
m
~ "~ --~ ~ ~ ~

~ ,~a
.~ ~m
i

' . ~
~ e

N~
"~ ~ ~

I
&
~ N

N
.N ~d
ul o
, 0

o
~ m
N
m31 --

rare in nature. A review of the literature failed to turn This is illustrated by the fact that I. Geoffroy St.
up any convincing case of a monster in which a single Hilaire's classification, as many other subsequent ones,
body axis trifurcates anteriorly into three heads. I. was not based on any particular species. Any vertebrate
Geoffroy St. Hilaire summarized the classical literature monster could be fitted into his scheme. For example,
and concluded that with only two exceptions all the few as mentionned above, bicephalous monsters are found
other reported cases are at best doubtful, and most in most vertebrate species (fig. 6). Similarly, a cyclops
Medieval descriptions of three headed animales are exhibits the same characteristic morphological featu-
clearly the products of fantasy (like the tricephalous res, regardless of the species in which it appears (Adel-
human reported by Bartholin ; born with three heads man 1936). Sometimes this morphological invariance
- one human, the other a wolf's an the third a bloody can transcend the vertebrate-invertebrate boundary ;
mass without skin - was not only born alive, but " i t " for example, Ranzi (1957) reports a case of cyclopia
only expired after appearing in front of the city Senate in the squid. I refer to this organizational invariance
to make a series of sinister predictions). Of the relia- across species boundaries as transpecific parallelism.
ble descriptions of three - headed monsters, the Italian It suggests that the rules that control the expression of
physicians Reina and Galvagni provide the best des- abnormal forms exist above any historical considera-
cription (Sopra un feto h u m a n o trieefalo (1832) cited tions associated with lineage relationships.
in I.G. St. Hilaire, p. 341 ; see figure in Gould & Pyle
1937, p. 167). Based on Reina and Galvagni's detailed This point can be further illustrated using the tho-
dissection, it is clear, however, that this specimen is only rough anatomical study of Wright & Wagner (1925).
an apparent trifurcation, since the fetus had two com- As part of a research p r o g r a m on the genetic basis of
plete spinal chords running parallel to each other. One morphological abnormalities, Wright & Wagner asso-
of the two b o d y axes bifurcated to generate the extra, ciates were able to study over 300 guinea pig mutants
third, head. The absence of triple monsters was also which exhibited a variety of malformations. Wright &
noted by one of the pionners of G e r m a n comparative Wagner (1925) assigned these teratologies to the family
anatomy and embryology, Meckel, who in the early of monsters which I. G e o f f r o y St. Hilaire had termed
1800's elevated this fact to the category of law. Mec- " o t o c e p h a l i e n " . The affected specimens could be
: kel claimed that monsters exhibiting extra elements arranged along a gradient of increasingly severe mal-
were always generated by binary duplication of parts formations belonging to the various types ( " g e n e r a " )
(cited in I. Geoffroy St. Hilaire 1836, p. 331-332). Stoc- described by Geoffroy St. Hilaire. The less affected spe-
kard (1921) reported on an extensive study of naturally cimens (grade 1) had a slightly shortened lower jaw with
and experimentally induced " d o u b l e m o n s t e r s " in the loss of the incisors. Increasingly malformed foetuses
fish Fundulus. Out of several hundred specimens, 150 included progressive reduction of head structures, such
double monsters were observed. Only one triplet was as loss of nostrils, the formation of a univentricular
generated, which consisted of a combination of one telencephalon, and associated cyclopia. The sequence
embryo and a double headed monster (i.e. analogous of loss of structures was, however, highly invariant and
to the case reported by Reina & Galvagni 1832). the immense majority of the specimens studied could
be classified along the series illustrated in fig. 7.
I only mention the absence of three-headed mor-
phologies, but the same kind of analysis could be The point relevant to this discussion, however, is
applied to any other syndrome or abnormality. For that the observed morphologies are not exclusive to gui-
example, I have recently discussed digital anomalies in nea pigs. Numerous cases of cyclopia, otocephaly and
domestic breeds of dogs (Alberch 1986) ; a case in anencephaly have been reported in the literature. The
which perhaps there are no strictly forbiden m o r p h o - morphological patterns described are essentially the
logies but where clear trends in the generation of varia- same in all vertebrates (see review by Adelmann 1936).
tion can be observed. Stockard (1921) pointed out that although the spe-
In conclusion, not all monstrosities are possible cific morphology of the malformation obtained is iden-
and a m o n g the possible, certain ones are more likely tical, there are species-specific differences in the pro-
to be generated than others. In addition, observed mor- bability of occurrence of any given teratology. In other
phologies can be arranged in unambigous graded words, certain teratologies are commonly encountered
sequences. Therefore, teratological f o r m m a y not in certain species while rare in others. For example,
necessarily be discrete but it is always orderly. double monsters are c o m m o n in the trout, while they
are a lot more difficult to obtain in the minnow Fun-
b. Transpecifie parallelism in the genesis of teratolo- dulus. Similarly, Stephens & Shepard (1983) reported
gical forms different incidences of specific limb teratologies in
various human ethnic groups. Similarly, it is well
Similar trends in the kinds of monstrosities gene- known that there is a hereditary component in the for-
rated are encountered in different, unrelated, species. mation of monozygotic twins in humans (e.g. Bulmer,
-- 32 --

0 d

,i -~..'~'..:":~:~';~;~iii.;:........-....T;7:'"::'.'::':':'::.::'?:"~ : ~
:~/:.!.:.~' " " ......... : ' ii'2:.!~,~-~-~~!~

, ~

C f

Fig. 6 - - Ordered variation and transpecific parallelism in the occurrence of monstrosities.


The three basic types of conjoined twins ("double monsters") in humans (a-c) and in the fish Fundulus (d-f). Duplication results
from anterior bifurcation of the body axis (a, d). Fully developped twins can " f u s e " through any region in their body but, most
frequently, fusion occurs in the abdominal region (b, e). Parasitic twins are also observed in all vertebrate taxa (e. g. c, f). Drawn
from photographs in Stockard (1921) (a, b, d, e, f) and Drimmer (1973) (c).
Variation ordonn~e et paraIWiisme transp~cifique de monstruosit~s.
Les trois types fondamentaux de jumeaux siamois ("monstres doubles") chez l'homme (a.-c) et chez le poisson Fundulus ~d - f).
La duplication r6sulte d'une bifurcation antdrieure de l'axe du corps (a, d). Des jumeaux compl~tement d~velopp6s peuvent iusion-
ner par n'importe quelle partie de leurs corps, mais le plus souvent la fusion concerne ta r6gion abdominale (b, e). Des jumeaux
parasites existent aussi dans tousles taxons de vert~br6s (c, 0- Dessins a, b, d, e, f d'apr~s des photographies de Stockard (1921) ;
dessin c d'apr6s Drimmer (1973).
--33 m

1970). Monozygotic twins, resulting from an abnormal colorful and, like most bestiaries of its kind, bizarre
fission of the early blastocysts, are conceptually homo- treatise "Monstres et Merveilles", first published in
logous to any other teratology (except in armadillos in 1573. Pard's classification of teratologies was, in part,
which polyembryony is the norm rather than be based on rudimentary notions of the mechanisms that
exception). control development. To Pard, normal development
was the result of mixing the correct amounts of male
The recurrence of similar teratologies in different
and female " s e e d " . A class of monsters would result
species suggests that the structure of developmental
interactions responsible for the generation of a parti- from failure to mix adequate amounts of the two ingre-
cular structure (e.g. vertebrate head or limb) has remai- dients. Thus, one could obtain monsters resulting from
ned invariant throughout phylogeny. The species - or " t o o great a quantity of seed", which would cause the
even population - specific differences in the probabi- formation of extra structures (e.g. conjointed twins or
lity of occurrence of a particular teratology indicates extra limbs), or, alternatively, abnormalities could be
that there is a genetic ( = lineage dependent) compo- generated by "lack in the quantity of seed" (e.g. cyclo-
nent that determines that, at a particular state, certain pia or limblessness). Another category of monsters -
transformations are more likely than others. These consistent with his scheme and beliefs of the time -
species-specific tendencies can be understood using the would be the ones resulting f r o m mixture or mingling
conceptual framework outlined in Oster & Alberch of seed from different species. Interspecific hybridiza-
(1982). The probability of a particular transformation tion was commonly accepted to occur at the time and
in determined by the position of the system in para- the literature is full of examples of half h u m a n / half
meter space (Appendix II). beast creatures having been born out of bestial sexual
relations. Pard had numerous other categories ranging
from plausible mechanical constraints in the womb to
2. I N T E R P R E T A T I O N OF T H E O R D E R OBSERVED IN hermaphrodites to examples of the wrath and glory of
TERATOLOGY God. I do not want to enter into the details of Pard's
A genetic mutation can a priori result in any kind classification (see Pallister 1982, for further informa-
of morphology. There is no justification at the genetic tion on Pard and his work), except to illustrate one of
level to explain the fact that a two-headed morphology the earliest examples of how a specific model of deve-
is of much more c o m m o n occurrence than a three- lopment was used to construct a classification of tera-
headed one. (Since they are both maladaptive, and tological form.
usually lethal, " m u t a t i o n s " , the effects of selection can
A much more sophisticated, and scientific, classi-
be ruled out as an explanation in this particular case).
fication of teratologies is offered in the treatise by Isi-
Similarly it is very unlikely that the so-called "cyclo-
dore Geoffroy St. Hilaire, already mentioned in the pre-
p i a " mutations encountered throughout the vertebra-
vious section. I. Geoffroy's classification draws exten-
tes have the same genetic identity, in the sense of invol-
sively f r o m the theories of his father Etienne and his
ving changes in the same region of the DNA. The fact
father's disciple, the medical anatomist Etienne Serres.
that different mutations can result in the same mor-
These authors belonged to the school of French Trans-
phological outcome has been clearly demonstrated. For
cendental morphology (Russell 1916), closely associa-
example, there is a variety of mutations resulting in win-
ted with German Naturphilosophie. French transcen-
glessness in chickens (Waters & Bywaters 1943 ; Lan-
dentalists shared a set of assumptions with their Ger-
caster 1968 ; Prahlad et alii 1979). Similarly, it has been
man colleagues. They included the belief that all ani-
shown that several homoeotic mutations in Drosophila
mals are built upon a single structural plan, the idea
can cause the same morphological transformation
of a "chain of being" (the Aristotelian concept that
(Lewis 1968 ; Garcia-Bellido 1977).
all animals can be arranged in a single graded scala
Teratologies are very often caused by genetic naturae according to their degree of perfection, as mea-
mutations, but the explanation of the resultant patterns sured by their structural complexity, see Lovejoy 1936)
of morphological organization has to be searched for and a belief in recapitulation (the concept that argues
at the developmental level. that animals, as they develop, go through a series of
embryonic stages that parallel the chain of being).
a. Brief historical o v e r v i e w
One of the most successful applications of the
Unlike evolutionary classifications based on theories of recapitulation was in the analysis of tera-
lineage relationships, teratologies have traditionally tologies (Gould 1977 ; p. 49-52). To the French trans-
been understood and classified on the basis of deve- cendentalists, as well as to the Naturphilosophers such
lopmental properties (Zwilling 1955). One of the ear- as Oken, Meckel, and to their descendants, the Haec-
liest semi-scientific classifications of teratologies was kelians, the ordered generation of teratologies was the
put forth by the French surgeon Ambroise Pard in his result of regulation within an invariant ontogenetic
-- 34 --

No

4ol 1 4b

5b

6o

9a ~ 9b
Fig. 7 - - Dorsal and lateral views of guinea pig skulls.
Normal (Na, Nb) and various degrees of head reduction, cyclopia and otocephaly (1 to 9) (from Wright & Wagner 1925). Drawing
arranged in increasing degree of reduction of head structures. Usually the lower jaw is the first to be shortened (1-2). In some cases,
however, the upper jaw is affected (1'). This latter stage probably should not be part of the otocephaly series.
Vues dorsale et lat~rales de cranes de cochons d'Inde.
Normal (Na, Nb) et diff6rents degr6s de r6duction de la t~te, cyclopie et otoc6phalie (1 b. 9) (d'apr~s Wright & Wagner 1925). Dessins
ordonn6s par ordre croissant de r6duction des structures de la t&e. Habituellement la mandibule est raccourcie en premier 0-2).
Dans certains cas cependant la mfichoire sup~rieure est affect6e (1'). Ce dernier stade n'existe probablement pas dans la s6rie
otoc6phalique.
m 35 m

sequence (reflected in the scale of being). For exam- ted conditions found in lower forms. The former,
ple, Oken emphasized the relationships between tera- however, did not result in an extension of the scale of
tological and foetal structure, and affirmed that " m a l - being into new designs. Instead, monsters by excess
formations are only persistent foetal conditions" (cited always resulted in duplication of existing structures.
in Russel 1916, p. 91). In a similar vein, I. G e o f f r o y
The recapitulatory ideas of early teratologists were
discusses the parallelism of the "serie zoologique" with
abandoned, even if their particular taxonomy has often
the "serie teratologique". He argues that teratologi-
been retained, by studies based on more dynamical
cal anomalies in advanced species reproduce the con-
models of development and pattern formation. Uni-
ditions normally found in " l o w e r " species. As in the
versal rules need to be replaced by system-specific inter-
case of some of the analyses of Par6, we are dealing
pretations, as illustrated in the next section.
with a system of classification of f o r m rooted on the
developmentalist concepts of the time. b. A d e v e l o p m e n t a l analysis o f head reduction and o f
Gould (1977) discusses a celebrated example of this c o n j o i n e d twins : a c o n t e m p o r a r y classification o f
type of interpretation of teratologies on the basis of monsters
recapitulation. Serres, through a detailed comparative The observed regularities in the generation of tera-
anatomical study claimed to have demonstrated that tological form can be explained in terms of morpho-
a headless (anencephalic) human fetus was a condition genetic properties. I will illustrate this point by exami-
generated through an arrest of development. In this ning two specific teratological systems: the trend
particular case, Serres argued that the embryo had stop- towards increasing loss of head structures and the gene-
ped developing at a stage in the "scale of being" homo- ration of conjoined twins.
logous to a mollusk.
My choice of these two seemingly unrelated clas-
A very influencial dichotomy drawn by I.G. St. ses of teratologies is not an arbitrary one. Several clas-
Hilaire was that distinguishing between monsters by sical authors had already pointed out that they may
excess and monsters by default. The latter simply reflec- reflect a single developmental trend (e.g. Wilder 1908

Fig. 8 -- "Cosmobiotic series" from cyclopia to axial duplication


from Wilder (1908).
Wilder defined Cosmobia as the set of all possible forms
that could be generated by continuous regulation in rates
of development.
"S6rie eosmobiotique" depuis un stade cyclopejusqu'~ une
duplication axiale d'apr6s Wilder (1908).
Wilder d6finit la "s6rie cosmobiotique" comme l'ensem-
ble de toutes les formes possibles qui pouvaient ~tre pro-
duites par une r6gulation continue des taux de
d6veloppement.
- - 36 - -

and Stockard 1921, fig. 8). These authors believed, on induction. The observed sequence of changes in mor-
the basis of comparative analysis and experimental per- phology (fig. 7) is the result of a combination of pri-
turbation, that the headlessness otocephaly cyclopia mary effects, such as lack of inductive events, and a
normal axial duplication sequence was the result of series of correlated changes in the patterns of growth
quantitative regulation of some underlying develop- and differentiation. For example, the loss of certain
mental process. Stockard (1921) specifically proposed cranial bones, such as the alisphenoids and palatines
a slowdown in the rate of development at various could be interpreted as being correlated with a shorte-
embryonic stages as the single cause of this sequence ning of the mandibular arch (which is the result of a
of form transformations. That a mechanistic relations- slow down in the rate of migration of neural crest cells
hip exists among these teratological conditions is sup- into the region). These feedback interactions between
ported by the recent research of Gerhart and collea- the central nervous system, sensory organs, chondro-
gues (Gerhart et alii 1986) who have convincingly shown cranium and dermatocranium have been well demons-
that both reduction and duplication of head structu- trated experimentally (e.g. see reviews by Murray 1936 ;
res can be generated by alterations of the process of Katz 1983).
cytoplasmic compartmentalization in the early cleavage
stages. This perturbation in early development results The results of Wright & Wagner (1925) illustrate
in changes in the patterns of mesodermal-ectodermal the highly dynamic and interactive nature of the deve-
interactions that characterize the establisment of the lopmental system. As they conclude (p. 439) : " T h e
body axis. case illustrates the extent to which morphological chan-
ges of an apparently qualitative sort may be brought
For example, cyclopia and related conditions about as a result of ramifying correlative effects from
shown in figure 7, can be explained in terms of a quan- initial gene effects of a simple quantitative s o r t " .
titative perturbation of a single developmental para-
meter : rate of mesodermal migration during gastru- Hyperdevelopment of anterior structures as well
lation. This migration rate in turn determines the as axial duplication can also be caused by changes in
amount of head mesoderm at the time of induction of the patterns of cytoplasmic compartmentalization
head structures (Adelmann 1936 ; Juriloff et alii 1985). during early development (Black & Gerhart 1986). We
Thus, a quantitative perturbation in a single parame- must distinguish this mechanism from other ways of
ter triggers an ordered series of qualitatively different generating conjoined twins. In fact, a classification of
morphological states. double monsters should be based on such mechanistic
The point to emphasize here is not the specific knowledge. Many early authors, such as I. G e o f f r o y
mechanisms controlling a particular process but the fact St. Hilaire (1836) or Stockard (1921) assumed that the
that although the final morphological transformation only way to make a double monsters was by the fusion
is highly invariant, the relationship between it and the (or failure to completely separate) of monozygotic
initial developmental perturbation is extremely com- twins. A review of the literature suggests that there are
plex. For example, let us assume that a mutation affects probably several different ways to generate a mamma-
the rate of polymerization of tubulin molecules into lian conjoined twin. The three varieties of incomple-
microtubules. This genetically mediated quantitative tely separated twins appear to be manifestations of dif-
change in the rate of microtubule assembly will affect ferent processes operating at different stages of deve-
the rate of cytoplasmic rearrangement that follows the lopment. The first class in composed of cases in which
fertilization of the egg (Vincent et alii 1986). This the inner cell mass fails to completely separate after an
change in the redistribution of cytoplasmic components initial fission (Hsu & Gonda 1980). Another possibi-
will not have an effect until the time of gastrulation. lity is a secondary fusion of twin blastocysts. In these
At gastrulation, changes in the cytoplasmic make-up two cases, the double embryo may be joined at any
of invaginating, anteriorly migrating, mesodermal cells region, the two parts will not be symmetrical (this is
will cause a reduction in the amount of head mesoderm probably the causal mechanism of the so-called para-
(Suzuki et alii 1984 ; Gerhart et alii 1986). A smaller sitic twin teratology illustrated in figures 6c and 6f),
population of head mesoderm cells will trigger a series and two amniotic cavities will be present. The speci-
of abnormal inductive events in the overlying ectoder- mens illustrated in figures 6b and 6e have most likely
mal neural plate (Adelmann 1936), as well as changes resulted from incomplete division of the early blastula
in the number of neural crest cells, which can have a stage. The third class of double monsters are the ones
wide range of secondary effects in the region (Noden resulting from a bifurcation of the primitive streak
1986). (Herring & Rowlatt 1981 ; Partlow et alii 1981 ; Eas-
ton 1985). This class of teratologies can be diagnosed
However, as already noted by Wright & Wagner in mammals on the basis of the presence of a shared
(1925), the final morphology is not only the result of single amniotic cavity, and the presence of a single axis
these fundamental changes in gastrulation and primary that branches anteriorly into two axes resulting in
m 37 - -

symmetrical anterior structures (fig. 6a and 6b). Black developmental parameter (e.g. number of mesodermal
& Gerhart's (1986) centrifugation experiments of X e n o - cells in the head region) above a certain threshold can
p u s eggs also result in conjoined twins with a bifurca- trigger a bifurcation of an anteriorly growing axis. Con-
ted body axis (see also Kunieda & Wakahara 1987). versely, a trifurcation is extermely unliked ; an unsta-
A developmental analysis of this sort is not only ble pattern requiring a very delicate choice of parame-
useful to classify teratological forms but would even- ter values. One may speculate that the rarity of triple
tually allow us to predict which morphologies are easier monsters, discussed by Meckel and I. Geoffroy St.
to generate. For exaniple, the pattern formation models Hilaire, may be a reflection of the inherent difficulty
discussed by Oster & Alberch (1982) and Oster et alii of causing a trifurcation of the anteriorly advancing
(1988) suggest that a smooth perturbation of a given primitive streak or archenteron during gastrulation.

C. A D E V E L O P M E N T A L A P P R O A C H TO T H E ANALYSIS OF E V O L U T I O N A R Y PATTERNS -
T H E A D A P T I V E R A D I A T I O N OF T H E VERTEBRATE LIMB

Limb diversity in tetrapods is a classic example of comparative analysis (Alberch & Gale 1985 ; Shubin
~,,~vtive diversification within a conserved body plan. & Alberch 1986) and experimental manipulation of
Consider the widely divergent functions and drastically development (Alberch & Gale 1983 and unpublished).
different external shapes of the wing of a bat, the leg The basic hypothesis is that "rules of construction"
of a horse and the fin of the whale. In spite of these determine not only how a limb is built during ontogeny
differences, all tetrapod limbs share a similar skeletal but also how it can be modified through phylogeny.
arrangement. The vertebrate limb is usually divided into It is the nature of these developmental properties which
a proximal segment or stylopodium (always composed is responsible for the observed invariances in the evo-
of one bone, the humerus in the forelimb and the femur lutionary diversification of tetrepod appendages.
in the hindlimb), followed by one containing two bones,
the zeugopodium (radius-ulna ; tibia-fibula), which, in Based on empirical observations, Shubin &
turn, is connected to the distalmost segment, the auto- Alberch (1986) proposed that limb skeletal patterning
podium, including a set of wrist bones followed by should not be viewed as the global specification of inde-
digits. It is astonishing that this morphological plan has pendant cartilage foci. Instead, the skeletal pattern of
been conserved for millions of years of evolutionary the limb results from an initially continuous cartilagi-
history even in the face of divergent selective pressures. nous condensation that grows anteriorly, and sequen-
tially branches and segments to give rise to discrete ele-
The vertebrate limb played an important role in ments. A single proximal element (femur/humerus)
the elaboration of transcendentalist (i.e., internalist) begins to grow distally and subsequently branches into
ideas. It was a key example to illustrate the ideas of two elements (tibia/fibula ; radius/ulna). Then, a more
early pre-evolutionist German natural philosophers complex pattern of branching events is responsible for
such as Oken and Goethe and subsequently adopted the genesis of the carpus/tarsus and digits. Individual
by such classical morphologists as Owen and Gegen- skeletal elements form by the splitting ("fission") of
bauer. The limb with its metameric organization and an initially continuous cartilaginous rod (fig. 9). The
conserved pattern of connections among skeletal ele- observations of Shubin & Alberch (1986) support a
ments was a perfect example of serial repetition of parts model of pattern specification in which local pattern
within an invariant body plan ; a much more unambi- emerges as the result of local morphogenetic interac-
guous example of the postulates of the Naturphiloso- tions ; they are consistent with the mathematical models
phie School than the widely cited but more controver- of pattern formation proposed by Oster et alii (1985)
sial claims about the segmental organization of the ver- and Oster et alii (1988). In fact, mathematical mode-
tebrate head proposed by the same authors. ling, comparative analysis and experimental perturba-
The externalist explanation of this pattern has been tions suggest that all limbs are constructed as variations
that limbs share a common structural plan because they in the iteration of three basic types of cartilage con-
are all derived from a common ancestor. Most aspects densation. A cartilage element can form de n o v o as an
of this plan have not been changed by selection because independant foci. It can branch and split into two ele-
is hat not been necessary. ments ( " b " event in figures 9 and 10) or, alternatively,
a continuous rod of cartilage can segment into two or
In this section, i briefly summarize recent work more elements (denoted as " s " events in figures 9 and
that addresses this problem from a renewed internalist 10). For example, figure 10a illustrates, superimposed
perspective. Such an approach requires the integration on a fully developped limb, the sequence of branching
of mathematical modeling (Oster et alii 1985 and 1988), and segmentation events that give rise to the skeleton
- - 38 - -

A d

~5

13

Fig. 9 (A) Low-power and (B) high-power photomicrographs showing the embryonic connections in the developing carpus of a larval Ambys-
toma which has been cleared and the chondrocytes stained with Alcian Blue.
(B) Illustrates the branching event ( " B " ) that produces the ulnare (u) and intermedium (i) and the segmentation event ( " S " ) that
produces the radiale (r) from the radius (R). In the low-power view (A), the overall pattern is seen. Notice the segmentation events
that produce the radiale (r) from the radius (R), element (y) from the radiale (r), and the centrale (c) from the intermedium (i).

Microphotographies ~ faible grossissement (A) et h fort grossissement (B) montrant les connections embryonnaires dans les carpes
en d~veloppement de larves d'Ambystoma qui ont ~t~ ~claircies et dont les chondrocytes ont ~t~ color,s au Bleu d'Alcine.
(B) montre l'6v6nement de bifurcation ( " B " ) qui donne l'ulnare (u) et l'intermedium (i), et l'6v6nement de segmentation ( " S " )
qui donne la radiale (r) ~t partir du radius (R). La rue ~ faible grossissement (A) donne le sch6ma d'ensemhle. A noter les 6v6nements
de segmentations qui donnent la radiale (r) ~t partir du radius (R), l'616ment (y) ~ partir de la radiale (r) et la centrale (c) ~ partir
de l'intermedium (i).
-- 39 --

of the generalized five-toed, limb of the salamender Nevertheless, close observation of these morphologies
A m b y s t o m a mexicanum. indicates that they are built as interactions of the three
Experimental manipulation of normal limb deve- basic types of cartilage condensation patterns shown in
lopment with colchicine, and other mitotic inhibitors fig. 9.
with reversible effects, can generate a variety of abnor- Shubin & Alberch (1986) have suggested that these
mal morphologies (Alberch & Gale, 1983, 1985; basic morphogenetic rules for skeletal differentiation
Alberch, in prep.). Alberch & Gale (1983) have argued have remained invariant throughout vertebrate history.
that changes in pattern formation occur as the domain That is, all vertebrate limbs result from a combination
of the early embryonic field (the size and number of of de novo, branching and segmentation events. Besi-
cells of the limb bud) is reduced below a certain thres- des these local morphogenetic interactions, globally
hold - a hypothesis consistent with mathematical specific asymmetries are also critical in controlling limb
models of pattern formation (e.g. Oster et alii 1988). patterning. Such " g l o b a l " properties are basically the
The important point is that the malformations indu- result of groups of specialized cells exerting long range
ced by the colchicine treatment are not chaotic ; to the influences on the vertebrate field (e.g. the apical ecto-
contrary, there is a high degree of order. As discussed dermal ridge or the zone of polarizing activity ; see
in previous sections, the ordered generation of terato- Hinchliffe & Johson 1980 for review). The evolution
logies is one of the fundamental properties of the mor- of limb diversity is the result both of changes in cellu-
phological systems. Figure 10b-c illustrate two stages lar morphogenetic properties that either affect local
in the series towards increasing reduction of digits and interactions, such as cell adhesiveness or extracellular
skeletal elements induced by colchicine treatment. This matrix constitution, or global parameters, such as the
is a true "cosmobiotic series" (fig. 8), in the sense that slope of a gradient o f morphogen concentration. In
the realm o f morphologies obtained through experi- both cases a perturbation in cellular parameters can
mental manipulation is the result of perturbation of one result in alterations in the sequence of branching and
or more underlying developmental parameters. The segmentation events. For example, salamanders appear
property o f transpecific parallelism is also illustrated to develop their limbs in ways quite different from other
in this example. For example, the teratologies illustra- teti:apods (see reviews in Shubin & Alberch 1986 and
ted in figures 10b-c correspond, to a striking degree of in press). In particular, the sequence of digit f o r m a -
similarity, to morphologies found in derived urodele tion appears to be reversed (e.g. Alberch & Gale 1983).
species (figures 10d-e). Alberch & Gale (1985) documented a parallel between
This transpecific parallelism can be explained in the experimentally generated patterns and the phylo-
terms of the experimental perturbation affecting the genetic trends towards digital reduction observed in
sequence of branching and segmentation chrondroge- frogs and salamanders. Reduction of digits, a pheno-
nic events. For example, the cave-dwelling salamander menon that has occurred independently several times
Proteus anguinus (fig. 10e) represents a highly derived in the evolution of amphibians, entails the loss of the
limb morphology characterized by an extreme case of innermost (preaxial) digit in frogs, but the outermost
loss of tarsal elements and digits. As illustrated by the (postaxiat) one in salamanders. When a generalized sala-
parallel between Figures 10c and 10e, the morphology mander limb bud was treated with a mitotic inhibitor
of Proteus is the result of a reduction in the number of a postaxial digit was often lost. If the same experiment
chondrogenic segmentation events as well as the switch was performed on a frog, the first digit was lost (fig. 11).
from a branching to a segmentation event off the fibula. This parallel between phylogenetic trends and
Not all laboratory induced malformations result experimental results (coupled with the observed diffe-
in forms observed in nature. But even these " u n n a t u - rences in the ontogenies of frog and salamander limbs
ral" constructions, are limited by the morphogenetic (Shubin & Alberch 1986) suggests that the developmen-
properties of the system. For example, Patou (1973) tal differences that have evolved in these groups deter-
removed and disaggregated the mesoderm from chick mine the variations that can arise and become fixed
and duck limb buds. The two cell populations were during evolution. In reference to Appendix II (fig. 19
mixed, reaggregated and packed into an ectodermal jac- one would say that the frog and salamander limb-
ket. As expected, this procedure resulted in highly a- patterning systems have different evolutionary poten-
bnormal, and'non-functional, skeletal patterns (fig. 12). tials, in terms of the variation that each can generate.

D. TOWARDS A T H E O R Y OF FORM

Evolutionary biology needs a theory of form. Cur- responding to externally determined selective pressures.
rent neo-Darwinian theory is essentially a blend of Men- The concepts of regulation and feedback that characte-
delian genetics and a populational theory of natural se- rize developmental systems are conspicuously absent
lection. The organism is viewed as a black box, passively from existing theoretical schemes (Wake et alii 1983).
n 4 0 m

Fig. 10 - - A. Normal foot morpho-


logy of the salamander
Ambystoma mexicanum.
Treatment of the embryo-
nic limb bud of A. mexi-
canum with the mitotic
inhibitor colchicine results
in the formation of terato-
logical limbs with fewer
skeletal elements (e.g.
v',,_ ],
four-toed (B) or two-toed
limbs (C). This array of
morphologies could be vie-
wed as a "cosmobiotic
series" (see fig. 8). The
pattern of experimental
variation is not r a n d o m
and the teratological forms
obtained often correspond
to the normal morpholo-
gies observed in evolutio-
narily derived species of
urodeles, such as the four-
toed, Hemidactylium scu-
tatum (D) and the two
toed, Proteus anguinus
(E). Dotted lines indicate
the pattern and types of ,
cartilage connections
during development, s =
I ~ $
segmentation of a cartilage
rod, b = branching of a
chondrogenetic foci (see
fig. 9). Loss of elements in
the developmentally per-
turbed or evolutionary
derived species are usually
the result of failure of the
-

embryonic chondrogenetic
foci to split ( " s e g m e n t " )
into two. The failure o f a
s e g m e n t a t i o n event to
occur is denoted by an
asterisk (from Alberch &
Gale 1985 and unpublished
data).

A. Morphologie normale de
la patte ehez la salamandre
Ambystoma mexicanum.
Traitement du bourgeon
de membre embryonnaire
de A. mexicanum avec un
inhibiteur mitotique (col-
chicine) qui aboutit A des
m e m b r e s t6ratologiques
poss~dant moins d'~l&
ments squelettiques :
membres ~ qnatre orteils
(B) ou ~t deux orteils (C).
!S ) 7,"1
:*-J-!~~oteus
Cette succession de mor-
phologies peut ~tre com-
prise c o m m e une "s6rie
cosmobiotique" (cf. fig.
8). Le mod61e des varia-
tions exp6rimentales n'est
pas al~atoires et les formes F T
t6ratologiques r6alis6es g T
correspondent souvent
des morphologies normales
rencontr6es dans des esp6ces parentes d'urodbles c o m m e l'Hemidactylium scutatum A quatre orteils (D) et le Proteus anguinus
deux orteils (E). Les lignes pointill6es indiquent les modes et les types de connections cartilagineuses pendant le d6veloppement.
s = segmentation d ' u n e barquette cartilagineuse, b = bifurcation chondrog6n6tique (cf. fig. 9). La perte des 616ments par pertur-
bation exp6rimentale du d6veloppement c o m m e par 6volution naturelle r6sulte d ' u n e incapacit6 & se diviser en deux des points
chondrog6n6tiques embryonnaires. L'absence d'6v6nement de segmentation est signal6e par une ast6risque (d'apr6s Alberch & Gale
1985 et donn6es in6dites).
--41 --

III IV Fig. 11 - - An experimental approach


to the study of phyiogene-
SALAMANDERS (7) tic trends.
FROGS (2)
A. Two anuran lineages
have lost toes indepen-
dently during phylogeny.
In both cases the first toe
("the t h u m b " ) is the one
that has been lost. In sala-
manders digital loss has
occurred in at least seven

I independant cases during


evolution. In all of the ins-
tances the forms have con-
verged in the loss of the
fifth toe. B. Identical treat-
A. Evolutionary trends ment of an embryonic limb
bud with a mitotic inhibi-
tor with reversible effects
results in the loss of pre-
axial digits in frogs while
Ill IV salamanders lose post-axial
elements. These differences

FROGS
\l//
80g
in experimentally generated
forms parallel observed
evolutionary trends (from
Alberch & Gale 1985).

Approche exp~rimentale de

0
III IV
[[ I'~tude des tendances
phylog~n~tiques.
A. Deux lignGes d'anoures
..I
0
-r-
eL
e~
0
IE
I ont perdu leurs orteils indd-
pendamment au cours de la
phylogenGse. Dans les deux
cas le premier orteil (le
pouce) est celui qui a ~t6
ml perdu. Chez les salaman-
<1
ev
I'--
LLI
TIBIA /I I,
i FIBULA
III IV
dres la rdduction digitale
s'est produite dans au
moins sept cas indGpen-
dants au cours de l'6volu-
7 tion. Oans tousles cas c'est
le cinquibme orteil qui a 6t6
perdu. B. Le m~me traite-
ment d ' u n bourgeon de

SALAMANDERS
[[ membre embryonnaire par
un inhibiteur mitotique /t
effets rGversibles aboutit ~t
la perte du doigt prG-axial

I chez les grenouilles et ~ la


perte des 616ments post-
axiaux chez les salaman-
dres. Ces diffdrences entre
les formes gdndrGes expdri-
mentalement sont similaires
[3. Experimentally generated morphologies aux tendances 6volutives
observ~es.

In this section, I will define the concept of form, racterize development rather than on specific genetic
not as a static entity but rather as a process. My final constitution.
conclusion is that a theory of form must be based on
the properties of the network of interactions that cha-
- - 4 2 - -

1. W H A T IS F O R M ? Form embodies all the relations of the organism and


expresses the whole and its internal organizing proper-
The developmental biologist Weiss (1955) equated ties. This definition, rooted is the Romantic ideas o f
organic form with frozen development. With this sta- Goethe, represents a tradition now absent from evolu-
tement, Weiss meant that form cannot be dissociated tionary biology (Lenoir 1984 ; Brady 1984 ; Webster
from process. Form is not a static entity defined by the & Goodwin 1982). It stresses the concepts of interac-
spatial position of components, but a dynamic process. tion, integration and regulation. To understand form

Fig. 1 2 - Various patterns produced


by Patou's disaggregation-
reaggregation experiments.
Notice that in each case the
final pattern is produced by
a combination of brala-
ching and segmentation
events (from Hinchliffe A
1977).

Diff6rents motifs obtenus


dans les exp6riences de
d6sagr6gation-r6agr6gation
de Paton.
On remarque que dans cha-
que cas le motif final est
obtenu par une c o m b i n a
son d'6v6nements de bi:fu
cation et de segmematic
(d'apr6s Hinchliffe 1977

[3
-- 43 --

and its rules of transformation we need to know the that either regulate the expression of other genes, or,
properties and nature of its generative rules. These rules in the case of products o f the so-called morphogenetic
encompass genes, development and function in a man- genes (e.g. Edelman 1985 ; 1986a, b), determine mor-
ner that cannot be dissociated. phogenetic properties such as extracellular matrix com-
position, cell adhesion, mitotic rate, diffusion constant,
2. GENERATIVE RULES - kinetic activity, etc. Morphogenesis is the result of com-
THE MAPPING OF GENES TO PHENOTYPE plex physico-chemical interactions at this level. One of
There are two ways to conceptualize the relation- the outcomes of morphogenetic processes is that the
ships between genes, development and phenotype (fig. spatial relationships a m o n g cell populations are alte-
13). Since the discovery of genes as the units of here- red. Often, a so-called inductive event occurs when
dity, the?e has been a tendency to view genes as the populations of cells with different developmental his-
determinants of form. Genes control developmental tories are suddenly juxtaposed as a result of a morphoge-
processes, which in turn, generate f o r m (fig, 13a). If netic process. Induction implies that the expression of
this hierarchical scheme were correct, both morpholo- some genes is repressed while a new set is turned " o n " .
gical evolution and development could be reduced to The implications of this cyclical/feedback scheme
purely genetic problems. This is reflected in the posi- drastically alter our perception of how complex mor-
tion of some Mendelian geneticists who view evolution phologies evolve. Development cannot be reduced to
as a change in gene frequencies, or, in the molecular a problem of gone expression, since gone expression
biologist's view of development, as a temporal and spa- itself is under epigenetic control. Therefore, although
tial sequence of gene expression. gone frequencies are a valid method for evolutionary
This depiction of genes and development as inde- " b o o k k e e p i n g " (see Wimsatt 1980 ; Sober & Lewon-
pendent levels is incorrect in the sense that genes do tin 1982), we cannot have a purely genetic theory of
not specify development, or even form, because gene morphological evolution. Theory of f o r m has to be
action itself is intimately linked to developmental inte- based on the global properties of the network of inte-
ractions. This interactive nature of developmental pro- ractions that characterize development (fig. 13b ; see
cesses is illustrated in fig. 13b. Genes make proteins appendices I and II).

A B
MOP"OLOGY I /
l- G \
\
\
\

CELL morphogenesiI sTISSUE


\

~,
PROPERTIES GEOMETRY t1
I
I
L)EVELOPMENTAL protein l induction / ; 1
PARAMETERS synthesis~
/
/
/
\
\
--f GENES ~-- 1
/

i/" / "" "" X\l


GE.ES [ ,, gene /
I
\\ regulation / /.

Fig. 13 - - A. Hierarchical scheme of the genesis of form with genes as the controlling agents. B. Cyclical view of development in w h i c h genes
are j u s t one step in the c h a i n o f interactions. D o t t e d lines i n d i c a t e the two k i n d s of r e g u l a t o r y l o o p s discussed in the text. 1 - cell
a u t o n o m o u s ; 2 - epigenetic.
A . Schema hi~rarchique de ia gen~se de la forme 06 los g~nes agissent comme agents de contr61e. B. Vue cyclique du d~veloppement
d a n s lequel los g~nes sont une s i m p l e ~tape d a n s une cha~ne d ' i n t e r a c t i o n s . Los lignes pointill~es i n d i q u e n t los deux types de r6gula-
t i o n discut~s dans le texte. 1 - a u t o n o m i e cellulaire ; 2 - ~pigen6se.
-- 44 --

This feedback relationship can be extended to the In this case, developmental regulation is the result of
level of interactions between form and function. The complex physico-chemical interactions that determine
form/function dichotomy is one of the oldest problems cell behavior and extracellular matrix composition. In
of morphology, certainly since the Cuvier and Etienne turn, morphogenetic interactions result in the expres-
Geoffroy Saint-Hilaire debates (Russell 1916 ; Appel sion of heterogenous spatial patterns among groups of
1987). Again, the relationship was viewed as a hierar- cells - a prerequisite for inductive events.
chical one with either form determining function or vice
versa. I believe, however, that later stages in develop- a. Gene regulation
ment could be characterized as a dialogue between form
and function. For example, the movement of embryos Control of cell autonomous development, such as
inside the egg or the mother's womb may be more than the expression of homoeotic genes in Drosophila (Stern
a spurious epiphenomenon. Movement is a requirement 1954) or trans-determination behavior (fig. 4a), is
for adequate integration of the nervous, muscular and mediated by intracellular networks of genetic interac-
skeletal system [for specific references on the effects tion. The process of insect segmentation is one o f the
of movement on joint formation and bone shape see best understood examples of such type of developmen-
Drachman & Sokoloff (1966), Hall (1969), Lanyon & tal regulation.
Rubin (1984) ; for the effects of function on the deve- Extensive genetic analysis in Drosophila has revea-
lopment of the nervous system see Katz & Lasek (1978), led that two classes of genes expressed in the zygotic
Katz et alii (1981) ; and for a general discussion on the genome are responsible for the formation of the seg-
evolutionary implications of this form - function feed- mented pattern characteristic of all insects. "Segmen-
back see Alberch (1982) and Katz (1983) for general tation genes" are necessary to establish the correct seg-
reviews and Meyer (1987) for recent experimental ment number and polarity (Nusslein-Volhard & Wies-
work]. Therefore, it is impossible to state that form chaus 1980 ; Hiromi et alii 1986). This first step in the
determines function or vice versa since they are inter- process of specification of segment number is media-
connected at the level of generative process. ted by a complex set of interactions between the pro-
ducts o f " segmentation genes" and maternally inheri-
3. M O R P H O G E N E T I C RULES VS. M O R P H O L O G I C A L PAT-
ted cytoplasmic determinants (e.g. Hiromi et alii 1985 ;
TERNS : THE NATURE OF REGULATORY INTERACTIONS
see Meinhardt's model discussed in Appendix I). This
In previous sections I have demonstrated that cer- process results in a spatially banded pattern of gene
tain patterns of morphological organization avxt trans- expression. The resultant banded pattern is not enco-
formation remain constant through long periods of ded in any of the segmentation genes ; it emerges as
evolutionary time. The existence of order in the for- the result of the interactions among the products of the
mation of teratological form (section B. 1), or the evi- various genes involved in segmentation and the mater-
dence in support of developmental rules of organiza- nally inherited cytoplasmic determinants.
tion and evolution in the vertebrate limb (section C) While segmentation genes specify the number of
are examples of such phenomena. These regularities in metameric units, homoeotic genes are required to assign
morphological pattern suggest a high degree of inva- the proper identity to each segment (Lewis 1978;
riance in the structure of the developmental processes Garcia-Bellido 1986 ; Seeger & Kaufman 1987). Analy-
schematically illustrated in figure 13b. In this section sis of double mutants suggests that the establishment
I briefly review some empirical evidence in support of of segment number (an "epigenetic" regulatory path-
this hypothesis. In particular, my basic thesis is that, way) is independent of the process of specification of
.if the structure of developmental interactions is highly the differences among segments (a "cell autonomous"
conserved in evolution, there should be a finite num- pathway).
ber of generative processes. Morphological diversity is
the result of the repeated iteration of the same rules Homoeotic genes are one of the best examples of
in different contexts. an "intracellular" regulatory pathway (indicated as 1
in fig. 13b). These genes appear to directly control the
Developmental interactions are mediated by gene expression of a series of "morphogenetic genes" (i.e.,
products. There are two distinct pathways through genes that mediate development via the "epigenetic path-
which genes can regulate developmental processes. I way" by regulating developmental properties such as cell
refer to these regulatory pathways as "intracellular" adhesion, cell motility, reaction rates of diffusible mor-
and "epigenetic" (fig. 13b). The intracellular regula- phogens, composition of the extra-cellular matrix, etc.).
tory pathway is characterized by genes that directly con- Garcia-Bellido (1986) refers to the homoeotic genes as
trol the expression of other genes - the products of such "selectors" and suggests that a particular developmen-
"regulatory" genes never exiting the cell. An epigene- tal pathway is genetically specified by a combination of
tic pathway is one in which gene products are trans- selector genes in the " o n " or " o f f " condition. Unlike
ported to the cell surface or the extracellular matrix. the highly specific "selector" genes, morphogenetic
--45 --

genes (equivalent to Garcia-Bellido's " r e a l i z a t o r " So far I have argued for a limited set of gene con-
genes) are general in their expression. That is, they trol operations and for a finite number of morphoge-
mediate a wide variety of morphogenetic and induc- netic molecules. This analysis of developmental rules
tive processes (see below). can be extended to more phenomenological levels such
as cell behavior and patterns of tissue induction. Wes-
Garcia-Bellido (1983 and 1986) has argued, based sels (1982) convincingly reviewed the existence of a very
on available data on the roles of segmentation and reduced set of morphogenetic cell behaviors (e.g. cell
homoeotic genes, that the discreteness of the phenoty- shape change, chemotaxis, haptotaxis, etc.) that are
pic space discussed in fig. 1 is a direct reflection of the repeatedly used in different processes of organogene-
possible genomic combinations of selector genes. That sis. Similarly, available evidence based on tissue recom-
is, since he believes that few genes control the selec- bination experiments also supports the view that the
tion of alternative major developmental pathways (e.g. repertoire of inductive interactions is limited and evo-
decision to f o r m a leg a n d / o r antenna) and that each lutionarily conserved. For example, Nieuwkoop et alii
selector gene can only exhibit two states, " o n " or (1985) show in a comprehensive review of early chor-
" o f f " , there is a finite number of possible developmen- date development that, despite the great differences of
tal pathways. This number is determined by all possi- morphogenesis among the various chordate orders, the
ble combinations of genomic states, each combination inductive interactions are remarkably similar. That is,
corresponding to a specific class of phenotypes. variation emerges as the result of changes in the con-
text within which the inductive and morphogenetic inte-
Although radical - and, at least for vertebrates, ractions are iterated. The structure of the interactions
most likely incomplete - Garcia-Bellido's hypothesis is has remained constant for long periods of evolutionary
very relevant in the sense that it emphasizes an inter- time. This conservation of developmental systems is
nalist approach to morphological evolution. In addi- also illustrated by the vertebrate limb. The " rules" of
tion, he documents a fact critical to my argument : that limb development have remained remarkably invariant
there is a finite number of genetic " o p e r a t i o n s " at the since at least the Devonian, and probably earlier
level of intracellular regulation. (Hinchliffe & Johnson 1980 ; Shubin & Alberch 1986).

b. Epigenetic regulation of morphogenesis and In conclusion, the rules of interaction that con-
pattern formation trol development are limited and general. The itera-
tion of these rules in different contexts can generate
It is simplistic to view development as a simple an enormous diversity of patterns. This conceptual
sequence of gene control switches. Even the models of scheme was discussed by O s t e r & Alberch (1982) in an
cell autonomous regulation cited above have an "epi- evolutionary context. As an illustrative example we sug-
genetic" component in the form of gradients or waves gested that a wide variety of epithelial derivatives such
of morphogens (e.g. Garcia-Bellido 1977 ; K a u f f m a n as feathers, scales, glands and teeth could be viewed
1977; Meinhardt 1984; Ingham & Martinez-Arias as variations within a single morphogenetic process.
1986). Morphological pattern results from the interac- Our example was based on simple observation of mor-
tion between intracellular networks of gene regulation phogenetic events such as folding of epithelial layers
and extracellular (epigenetic) morphogenetic interac- of cells (Odell et alii 1981). Edelman et alii (1986) have
tions. Whatever order emerges f r o m the structure of recently proposed a specific model of epithelial mor-
networks of intracellular gene regulation is enhanced phogenesis based on their data on CAMs that integra-
by the existence of a recurrent and finite set of rules tes gene action, mediated by morphogenetic molecu-
of interaction that characterize the epigenetic regula- les, with morphogenesis (see Appendix 1). Burke (1988
tory. This view is supported by available data sugges- and in p r e p . ) using this conceptual perspective has spe-
ting that processes of organogenesis and pattern for- culated that a wide variety of structures, such as hairs,
mation are mediated by a relatively small number of glands, feathers, scales and even limbs and the cara-
molecules acting in a wide variety of contexts. pace of turtles, can be viewed as variations within a
single system of epithelial-mesenchymal interactions
This point has been elegantly illustrated by the (fig. 14). This scheme is, perhaps, analogous to Wil-
work of Edelman and colleagues on cell adhesion mole- der's cosmobiotic series, in the sense of these structu-
cules (CAMs) (Edelman 1985 ; 1986a, b), as well as res being ontogenetic and phylogenetic expressions of
with the work on a few other morphogenetic molecu- variations within a single developmental system. The
les such as fibronectin (Thiery et alii 1985) and cyto- key issue here is that morphological nevelty is not the
tactin (Grumet et alii 1985 ; Crossin et alii 1986 ; Mac- result of addition of new genes but of the iteration of
kie et alii 1987). This conceptual framework argues that the same rules in different contexts (see Nijhout et alii
there are few genes that mediate morphogenetic and 1986 for treatment of this conceptual issue using com-
inductive events. puter simulations).
- - 46 - -

4. R E C U R R E N C E OF SIMILAR M O R P H O L O G I E S IN DIFFE- If the tracks on which a developmental change can run


RENT TAXA IS A REFLECTION OF THE INVARIANT AND are prescribed and limited, quite different primary cau-
FINITE STRUCTURE OF DEVELOPMENTAL INTERACTIONS ses m a y lead to an identical effect : A railroad car will
roll along its track whether pushed by a gang of men,
Empirical data supports the view that morpholo- a locomotive, a small explosive charge, a vagrant breeze
gical transformations are the result of regulation within or whatever. I f the push is too great, it will, of course,
an invariant set of interactions rather than the outcome be derailed."
of the addition of qualitatively new genetic informa-
tion. I f this is the case, transformations are not para- The internalist approach focuses on the determi-
meter specific ; changes in a wide variety of parame- nation of the " t r a c k s " available for the railroad car
ters can result in identical transformations (Appendix to move on rather than on the forces that fuel the move-
II). Therefore, convergences and parallelisms should ment. F r o m an evolutionary point of view, only per-
be commonplace in evolution, since they are a reflec- turbations based on genetic mutation are inherited - an
tion of the internal order and potentially of the system. argument often used to base the study of morphologi-
cal evolution at the level of genes to the exclus.ion of
Again this property can be illustrated both at the any other levels of interaction. However, if we want
level of intracellular and epigenetic controls of deve- to understand the organization of the generative
lopment. For example, data f r o m transdetermination system, we can gain insights by studying the results of
(fig. 4a) and homoeotic mutations indicate that certain perturbations at any o f the levels of interaction illus-
states are more stable than others and certain transfor- trated in figure 13b. In fact, one should expect paral-
mations more likely. For example, the first thoracic seg- lel outcomes. For example, returning to G e r h a r t ' s
ment appears to be a " s i n k " - a default option when model of development of the vertebrate body axis
cell homoeotic genes are turned " o f f " . The relevance (Gerhart et alii, 1986), an ordered reduction and loss
of the genetic control of segment specification in the of head structures can be induced at various levels of
evolution of insects has recently been discussed by Raft development. It can result by a genetic mutation that
& K a u f f m a n (1983). alters the structure of tubulin units and affects their
rate of polymerization - in fact, one can envision a wide
This lack of morphological "specifity" at this variety of mutations at different loci resulting in an
most basic level of gene control of development was identical physiological perturbation (e.g. see Wright's
already emphasized by Goldschmidt (1940) ; in his dis- (1934) genetic analysis of head reduction in guinea
cussion of the evolutionary implications of phenoco- pigs). The same morphology can be generated by a per-
pies, he writes : turbation in the physico-chemical rates of redistribu-
tion of cytoplasmic determinants after fertilization (e.g.
" M u t a n t and phenocopy, then, look alike because by U.V. irradiation or cold temperature shock). Also,
changed genetic action as well as action by a phenoco- if the model is correct, a perturbation at the tissue level,
pic agent is limited to definite tracks. There can be no such as the surgical removal of prospective mesoder-
doubt that in the last analysis the primary change pro- mal cells prior to gastrulation, should also result in the
duced both by phenocopic agent and mutated heredi- loss of head structures. In every case, the morphologi-
tary material must be of a chemical nature. But there cal outcome is identical. A cyclops will always have the
is no reason to assume that it is identical in both cases. nostrils above the single eye.

E. A S Y N T H E T I C V I E W

An internalist approach to morphological evolu- in a historically dependent set of "potentialities" [e.g.


tion can provide insights into the nature of evolutio- the different evolutionary tendencies towards digital
nary processes that could not have been obtained using reduction in frogs vs. salamanders (section C ; Alberch
the externalist approach centered around the concepts & Gale 1985]. Every stage in evolution has an intrinsi-
of selection and adaptation. I conclude, however, by cally determined set of possible transformations. After
pointing out again that order in evolution is a combi- every phylogenetic transformation the potentiality of
nation of deterministic agents at two levels : internally the system m a y be altered (appendix II). Selection, by
generated order based on the internal dynamics of deve- determining the probability of survival of each of the
lopment, and natural selection, dependent on the pro- possible transformations, indirectly affects the inter-
perties of organism-environment interaction. nal state of the species.
Internal factors and external forces are intimately
linked. By selecting on a particular phenotype, the deve- Most people would accept this synthetic scheme,
lopmental system is indirectly affected and this results reducing the problem to one of relative importance of
-- 47 --

INITIAL SECONDARY
INDUCTION INDUCTIONS

>

Epithelium

Morphogenesis
J d
e
>
Mesenchyme

g
/

Fig. 1 4 - Interactions between epithelial and mesenchymal populations of cells can result in a wide variety of organs such as hair (a), salivary
glands (b), teeth (c), feathers (d), scales (e, f), limbs (g), turtle carapaces (h) among others. These various organs can be classified
according to the developmental events involved in their genesis. They fill begin with an initial induction that triggers the formation
of an epithelial thickening and a mesenchymal condensation. This inductive event is followed by a series of morphogenetic foldings
of the epithelial layer. The epithelial layer can either invaginate or evaginate (Odell et alii 1981). Morphogenesis is followed by
a series of secondary inductions that trigger tissue specific gene expression. A further division can be established in organ systems
in which the evagination is followed by an epithelial outgrowth that draws mesenchymal cells into the epithelial sac. Examples of
such type of development can be observed in the limb bud(g) and the turtle carapace (h). (Adapted from Burke 1988).

Des interactions entre des populations de cellules 6pithdliales et m6senchymateuses peuvent aboutir/l une grande varidt6 d'organes
comme les cheveux (a), les glandes salivaires (b), les dents (c), les plumes (d), les ~cailles (e, f), les membres (g), les carapaces de
tortues (h) parmi d'autres. Ces organes varies peuvent ~tre class6s selon les 6v6nements du d6veloppement impliqu~s dans leur gen~se.
Ils commencent tous par une induction initiale qui d~clanche la formation d'un ~paississement 6pith61ial et d'une condensation m~senchy-
rnateuse. Cet 6v6nement inductif est suivi par une s6rie de plissements morphog~n&iques de l'assise 6pith61iale. Celle-ci peut soit
s'invaginer soit se d~vaginer (Odell et alii 1981). La morphogen~se se poursuit par une s6rie d'inductions secondaires qui d~termi-
nent l'expression g~nique sp~cifique des tissus. Une division suppl~mcntaire peut ~tre 6tablie pour les syst~mes organiques chez les-
quels apr~s l'6vagination, l'excroissance de l'@ith~lium attire les cellules m6senchymateuses dans le sac 6pith61ial. Les exemples
de ce type de d~veloppement peuvent ~tre observes pour les bourgeons de membres (g) et les carapaces de tortues (h). (Adapt~ d'apr~s
Burke 1988).
-- 48 --

internal constraint vs. natural selection. Traditionally, one. Nevertheless, since there are limits to how much
internal factors have been considered to play a relative complexity can be included in any given model while
to the powers of natural selection. This preference maintaining its predictive value, the dichotomy is use-
clearly depends on the kinds of questions in which one ful since it discriminates between two qualitatively dis-
is interested. I could rhetorically respond, with the often tinct research programs. I would like to suggest that,
used argument that the role of selection is secondary, at this stage in the development of the field, it is more
since the realm of possible phenotypes is internally useful to study the organization and evolution of form
determined, a fact that reduces the role of selection to f r o m the internalist perspective. This approach, based
a negative agent, i.e. one that only eliminates less adap- on an understanding of development, naturally incor-
Led forms. This argument, although correct, would be porates the phenomena of regulation, integration and
as misleading as the suggestion that selection can effec- constraint that are inherent to the concept of organism
tively achieve anything. (i.e. form) and which are absent in the conceptual fra-
From this perspective, it is obvious that the argu- mework provided by the externalist approach which
ment of selection vs. internal contraint is not a valid tends to view the phenotype as a static entity.

CONCLUSION

This paper has been an attempt to outline an ope- nalist approach - the search for rules of interaction and
rational approach to the integration of development transformation. Evolutionary processes m a y or may
and evolution as well as to briefly explore some of its not be continuous, but the internal forces that fuel the
conceptual foundations. As F u t u y a m a (1986, p. 440) process are few and discrete. It is because of this pro-
states in his recent textbook of evolution : " T h e impor- perty that I hope that a theory that studies f o r m not
tance of development and historical contingency has in terms of historical contingency and functional
always been recognized, but until recently it has been demands but rather on the basis of the properties of
the kind of formal, polite recognition accorded to a its internal rules of assembly is possible. Monsters are
stranger at an otherwise intimate p a r t y " . In my view a good starting point towards such a goal ; they repre-
the reason why development has not been integrated sent forms which lack adaptive function while preser-
into the existing corpus o f evolutionary theory is not ving structural order. There is an internal logic to the
a technical one (the " w e do not know enough about genesis and transformation of such morphologies and
development" type of arguments) but a philosophical in that logic we may learn about the constraints on the
one. To successfully study the role of development in normal.
evolution, one needs to replace the orthodox school of Acknowledgements
thought, dating back to Aristotle, of nature as a con-
tinuum, which has become entrenched into the Darwi- This paper was written while the author was
nian theory of evolution (Lovejoy 1936), with an alter- Fellow-in-Residence at the Neurosciences Institute of
native philosophy, whose roots can be found perhaps the Neurosciences Research P r o g r a m in New-York. i
in Hegel - a theory based on a more dialectical and dis- thank Paul Elias for comments, Emily Gale for tech-
continuous view of nature. The latter approach empha- nical assistance and Lazlo Meszoly for the artwork.
sizes the interactive aspects of natural processes and Support was provided by The Simon Guggenheim
studies them on the basis of the constraints emerging Foundation and NSF grants BSR 84-07437 and DCB
from such interactions. This is the essence of the inter- 86-09073.

APPENDIX I

Examples of networks of "developmental interactions" theory, be described in mathematical terms. One of the
simplest cases of "kinetic reaction" involves two "well-
At various points throughout this paper, I have stirred" (i.e. uniformly distributed troughout the reac-
invoked the properties of " t h e structure of interac- tion volume) chemicals, A and B, which react with each
tions" that characterize developmental systems. This other in a reversible manner. This reaction can be repre-
is an abstract concept that I would like to make more sented as,
precise by providing a few examples from selected refe-
rences. The purpose of this appendix is not to review kl
the models. For information on the specific details of A ~ B
the models I refer the reader to the original papers. k2
where k I and k2 denote the rate constants for the for-
Development is basically a very complex sequence ward and back reactions. If the two products react in
of biochemical reactions, which could, at least in a linear fashion the equation of motion for this system
-- 49 m

is then, d A / d t = k2B - kiA and similarly for B. Thus, substrate (activator-inhibitor) manner (fig. 15). Mein-
we can describe the rules of chemical interaction in the hardt (1982) has extensively reviewed this kind of
form of rate equations. Once the rules are known and diffusion-reaction models and Oster et alii (1988) have
cast in mathematical form, standard analysis can be recently reviewed the basic types of pattern generation
used to investigate their stability properties, phase- mechanisms and discussed some of its evolutionary
transitions and distribution of steady states (see appen- implications.
dix II). It is important to realize, however, that these
properties are not encoded in the variables, A and B, Meinhardt (1984 and 1986), also using an
nor in parameters such as the rate constants, rather they activator-inhibitor system of interactions, models the
are a property of the form of the dynamical equation. segmentation and compartmentalization events that
The structure of the mathematical equation describes characterize early insect development (fig. 16 a-b). As
the nature of the interactions and determines proper- I discuss in section D. 3, advances in molecular bio-
ties such as stability, and transition probabilities which logy make it possible to identify some of the gene pro-
are the subject of this paper. ducts involved in these interactions (Meinhardt 1986).
Note, however, that the properties of the activator-
The equation given above was selected because of inhibitor system of interactions are not encoded in the
its simplicity. It is not, however, very representative of genome. That is, the segmentation pattern is not enco-
real developmental processes. Most developmental ded in the so called segmentation genes ; it emerges as
systems are not "well-stirred" systems, rather one of the result of the interactions illustrated in figure 16 a.
the basic properties of development is the generation
of spatially heterogeneous distributions of molecules, Finally, Gatlin et alii (1986) have proposed a model
cells and tissue types. The vehicle for dealing with the integrating gene activation and morphogenesis to
dynamics of such spatially distributed systems is the explain feather pattern formation. The model is based
theory of partial differential equations in which the on the known properties of L-CAM, a glycoprotein that
equations of motion become field equations as they are mediates cell-cell adhesion. Their model depicts the
known in physics. Most mathematical models descri- interactions between the dermis and epidermis during
bing pattern formation during development are deri- the process of pattern formation (fig. 16 c-d). The para-
ved from the notion, originally proposed by Turing meters of the model were experimentally perturbed
(1952), that spatial heterogeneity can emerge from an using antibodies and the perturbations numerically
originally homogeneous state by means of diffusion- simulated. The agreement between the experimentally
driven instabilities. Basically, these models involve a generated morphologies and the numerically simulated
set of two or more chemicals that diffuse at different patterns support the validity of the hypothesized
rates and react with each other in a specific enzyme- pattern-generating interactions.

A P P E N D I X II
Parameter space and transformational diagrams: If we know the form of the pattern-generating
the developmental basis of phenotypic stability and function, we can construct a diagram in which for every
ordered evolutionary transformation combination of parameter values we have a correspon-
Parameter space
ding phenotype, such a diagram is known as parame-
ter space (fig. 17). This figure illustrates a hypotheti-
Theory of pattern formation centers around the cal parameter space composed of six phenotypes, A,
concept of a particular phenotype, P, emerging as the B, C, D, E, F determined by the developmental inte-
result of a series of temporal and spatial interactions ractions of two parameters x~ and x2. There are seve-
during development (appendix 1). These interactions ral general conclusions about the properties of pattern
are regulated by a series of genetically controlled mor- formation models than can be illustrated using this
phogenetic parameters, (xi, i = 1,2 .... m), which can figure.
either be molecular properties such as diffusion and
kinetic rates, or phenomenological variables, such as 1. Many combinations of parameter values will
elastic properties of the extracellular matrix, cell moti- result in the same phenotype, i.e. there is no one-to-
lity rates, degree of cell adhesion, etc. In general, the one correlation between genetically or environmentally
relationship between the morphogenetic parameters and mediated changes in parameter values and phenotypic
the phenotype can be mathematically stated as, transformation.
p = f (xi) 2. The stability of a particular phenotype is
where f is an unspecified function describing the nature directly related to the area (volume, if more than two
of the interactions (see references in appendix 1, for dimensions are involved) of its domain in parameter
specific examples of functional forms), while x i is a space. A larger domain implies that a broader range of
finite number of interacting morphogenetic parameters. parameter values will result in an identical phenotype.
-- 50 --

Fig. 1 5 - Schematic depiction of a diffusion-reaction model.


Two molecules, an activator and an inhibitor diffuse at different rates. The
inhibitor diffuses faster than the activator. The activator exhibits autoca-
Activator ir talysis, i.e., its rate of production at a particular point in space is propor-
tional to the amount already present and it is destroyed at a rate proportio-
SLOW nal to the amount of inhibitor present. The inhibitor, in turn, is synthesi-
zed at a rate proportional to the amount of activator present. This system
DIFFUSION exhibits the property of local activation and lateral inhibition required to
generate stable spatial patterns.
Description sch~matique d'un module de diffusion-r~action.
Deux mol6cules, un activateur et un inhibiteur diffusent Adiff6rentes vites-
ses. L'inhibiteur diffuse plus rite que l'activateur. L'activateur far de l'au-
tocatalyse, c'est-A-dire que son taux de production b. un point particulier
de l'espace est proportionnel ~tla quantit~ d6j~t pr6sente et qu'il est d~truit
/t une vitesse proportionnelle ~tla quantit6 d'inhibiteur pr6sente. En revan-
che t'inhibiteur est synth&is6 ~tun taux proportionnel a la quantit6 d'acti-
Inhibitor v vateur pr6sente. Ce systbmeposs~de les propri6t6s d'activation locale et d'in-
hibition lat6rale, requises pour g6n6rer des motifs spatiaux stables.
FAST
DIFFUSION

Fig. 1 6 - Two models of pattern formation illustrating how a system of interactions


can generate stable spatial patterns.
A and B : Diagram of molecular interactions (A) proposed by Meinhardt
(1984) to explain the generation of primary intrasegmental pattern forma-
tion in insects (B). It is assumed that the initial step in the determination
of segments is the formation of a periodic pattern of three different ceil
types / S, A, P. A juxtaposition of S and P leads to the formation of a
segment border. The diagram in (A) proposes that a spatial sequence of
three or more structures can be generatedif three autocatalytic feedback
A B loops exist (S, A, P) that locally) compete with each other (e.g. via a com-
mon inhibitor R) but support each other over a long range. Line indicates
short-ranging substances, dotted lines depict long-ranging interactions. C
and D : Computer model of feather pattern formation proposed by Gallin
et alii (1986). (C) Epidermal cells (L) linked by the cell adhesion molecule,
/X\ I-- l
/I
L-CAM, produce a signal (Es). Dermal cells in mesenchyme (R) res-
\\ // pond to E s by increased mitosis (M), by production of another
cell adhesion molecule, N-CAM (N), and by production of ano-
- .+ ther signal (Ds). D s induces placode formation (P) and down regu-
lates epidermal production of E s. Graph at right shows plots of
the production of E s as a function of the level of D s. (D) Spatial
pattern of mesodermal cell condensation that accompanies feather
pattern formation. This pattern can be generated by the model out-
lined in (C).
C L .......... -+- . . . . P 1.0I\\ Deux categories de modes de formation illustrant comment un
syst~me d'interactions peut g~n~rer des motifs spatiaux stables.
A et B : diagramme d'interactions mol6culaires (A) propos6 par
Meinhardt (1984) pour expliquer la fabrication des motifs intra-
R .............. %-~ N
segmentaux primaires chez les insectes (B). I1 est suppos6 que l'6tape
initiale dans la d6termination des segments correspond ~ la for-

I"'
Ds mation d'un motif p6riodique de trois types cellulaires diff6rents
(S, A, P). La juxtaposition de (S) et (P) conduit ~t la formation
d'une limite de segment. Le diagramme (A) propose qu'une
s6quence spatiale de trois structures, ou plus, peut ~tre g6n6r6e s'il
existe trois boucles de "feedback" autocatalytiques (S, A, P) qui
localement sont en comp6tition les unes avec les autres (e. g. via
un inhibiteur comme R), mais s'entrainent mutuellement ~. long
o~o~,oa terme. Les traits pleins indiquent les effets A court terme, les poin-
,,,o;,o;o; till6s d6crivent les interactions ~ long terme. C et D : mod61isa-
"O'O"O"" tion du mode de formation des plumes propos6 par GaUin et alii
o.O~O~O (1986). C. Les cellules de l'6piderme (L) li6es par une mol6cule
d'adh6sion cellulaire, L-CAM, produisent un signal (Es). Les cel-
_~O=O=O lules du derme dans le m6senchyme (R) r6pondent ~ Es en aug-
WidOW mentant la mitose (M), en prodnisant une autre mol6cule d'adh6-
sion cellulaire, N-CAM (N), et en produisant un autre signal (Ds).
Ds induit la formation de placodes (P) et limite la production de
Es par l'6piderme. La courbe de droite sch6matise la production
de Es en fonction du niveau de Ds. D. Motif spatial des massifs
cellulaires m6sodermiques qui accompagnent la formation des plu-
mes. Ce motif peut 8tre obtenu ~t partir du mod6le expliqu6 en C.
m51 m

For example, phenotype B, which has a relatively small bits phenotype D (which per se is a very stable one).
domain should, ontogenetically and phylogenetically, be However, the distribution of parameter values for the
less stable than, say, phenotype D. Thus, the area of members of species 1 places them near a bifurcation
a domain in parameter space is equivalent to Wadding- boundary (fig. 17). Any small perturbation that would
ton's concept of canalization (e.g. Waddington 1957). increase the values of one, or both, of the morphoge-
netic parameter values would trigger a phenotypic
3. The lines drawn in figure 17 correspond to sets transformation from D to either E or F. One would
of critical (xl, x2) values. They correspond to trans- predict the phenotype of this species to be fairly uns-
formational boundaries among phenotypes, i.e, a small table, in the sense that polymorphisms should be gene-
perturbation across the threshold value will result in rated at relatively high frequencies. Conversely, spe-
a qualitative change in phenotype. These boundaries cies 2 has a more stable phenotype A due to its solid
are known, in the jargon of dynamical system theory, position in the center of the domain (see section B. 1
as bifurcation boundaries (see Oster & Alberch 1982 for empirical examples o f these properties).
for additionnal discussion on this issue). Examination
of bifurcation boundaries suggests that many different Transformational diagrams
perturbations in both parameter values can result in an
identical phenotypic transformation. From the structure o f a specific parameter space
(e.g. fig. 17), we can derive information about the rela-
This property relates to the discussion in section tive probabilities of occurrence of specific phenotypes
D. 4, on the "genetic non-specificity" of phenotypic as well as about the most probable pathways of trans-
transformation. The structure of the bifurcation boun- formation. This analysis is completely independent of
daries in parameter space is the evolutionary relevant any considerations about the relative adaptation of the
property (the railroad tracks in Goldschmidt's quote possible phenotypes.
(p. 45 or Waddington's chreods), not the specific per-
turbation pushing the system over the threshold. In general, one would expect the most stable
phenotypes (i.e., the ones with broader domains in
4. The stability of a particular population of phe- parameter space) to be the most probable. For exam-
notypes will depend on its position in parameter space. ple, phenotype D is more likely to evolve than B in
For example, let us assume a species (Sp. 1) which exhi- figure 17.

C F

Fig. 17 -- Parameter space. X


Espace param~trC
E

\
X!
- - 52 --

.X" Fig. 18 - - Transformational diagram corresponding to the parameter

space illustrated in fig. 17.


Letters correspond to phenotypes, while arrows depict pos-
sible transformations.
Diagramme de transformation correspondant/l l'espace
paramdtrd de la fig. 17.
Les lettres indiquent des ph~notypes tandis que les fl~ches
d~crivent les transformations possibles.

lspecies I I Isp~ci~ 21

e\ /,,e
~ . ,,%
/ e e
Fig. 19 --Species-specific transformational diagrams illustrating the kinds of phenotypic variations that is
likely to appear in species 1 (sp. 1), characterized by phenotype D and species 2 (sp. 2) characteri-
zed by phenotype A (from fig. 17). Thicker arrows represent transformational pathways more pro-
bable than the ones depicted by thin arrows.

Diagrammes de transformation illustrant les categories de variation ph6notypique qu'il est vrai-
semblable de voir appara~tre dans l'esp~ce 1 (sp. 1), caract@is6e par le ph~notype D et dans l'es-
p~ce 2 (sp. 2) caract@is6e par le ph6notype A (d'apr~s fig. 17). Les fl~ches 6paisses repr~sentent
les itin6raires de transformation les plus probables.
- - 53 - -

This fact does not m e a n t h a t D wilt b e c o m e evo- s h o w n in figure 4. F o r e x a m p l e , s m o o t h p e r t u r b a t i o n


l u t i o n a r i l y " f i x e d " as a species. It c o u l d be t h a t B is o f x I a n d x2 in t h e d o m a i n o f p h e n o t y p e A c a n o n l y
a highly a d a p t i v e p h e n o t y p e a n d D c o r r e s p o n d s to a result in a t r a n s f o r m a t i o n to B o r D, with the p r o b a b i -
lethal t e r a t o l o g y . Then, in spite o f the r e p e a t e d a p p e a - lity o f going f r o m A to D being a lot higher t h a n t h e
rance o f p h e n o t y p e D in the p o p u l a t i o n , selection will trans f o r m a t i o n is p r o p o r t i o n a l to the length o f the
push the s y s t e m t o w a r d s B a n d m a k e it the n o r m . This b i f u r c a t i o n b o u n d a r y b e t w e e n the two p h e n o t y p i c
hypothetical scenario illustrates the interaction between domains.
c o n t r a v e n i n g internal a n d external forces in e v o l u t i o n
(fig. 18). E a c h species, or p o p u l a t i o n , will have a u n i q u e
If we a s s u m e t h a t genetic m u t a t i o n s result in rela- t r a n s f o r m a t i o n a l d i a g r a m d e p e n d e n t on its p o s i t i o n in
tively small q u a n t i t a t i v e changes in m o r p h o g e n e t i c p a r a m e t e r space. In figure 19, i have illustrated the
p a r a m e t e r values, we can further our analysis a n d trans- t r a n s f o r m a t i o n a l d i a g r a m s for species 1 a n d 2 in this
f o r m the p a r a m e t e r space, c o r r e s p o n d i n g t o P = f hypothetical example.
(xl, x2) in figure 17), into a k i n e m a t i c g r a p h d e p i c t i n g
all possible p a t h w a y s o f t r a n s f o r m a t i o n s a m o n g phe- The qualitative differences between flogs and sala-
n0types (fig. 18). I refer to this g r a p h as a t r a n s f o r m a - m a n d e r s in the p o t e n t i a l i t y to lose specific digits (sec-
tional d i a g r a m . This t h e o r e t i c a l c o n s t r u c t is c o n c e p - t i o n C) is p r o b a b l y an e m p i r i c a l e x a m p l e o f the t h e o -
tually a n a l o g o u s to the e m p i r i c a l l y derived g r a p h s retical d i a g r a m in figure 19.

BIBLIOGRAPHY

ADELMAN H.B. (1936) - - The problem of cyclopia. Quart. BRADY R. (1984) - - The causal dimension of Goethe's mor-
Rev. Biol., 11 (2) : 161-182 ; 284-304. phology. J. Social Biol. Struct., 7 : 325-344.
AGUR Z. & KERSBERG M. (1947) - - The emergence of phe-
BULMER M.G. (1970) - - The biology of Twinning in Man.
notypic novelties through progressive genetic change.
Amer. Nat., 129 (6) : 862-875. Clarendon Press, Oxford.

ALBERCH P. ( 1 9 8 0 ) - Ontogenesis and morphological BURKE A.-C. (1988) - - Epithelial-mesenchymal interactions


diversification. Amer. Zool., 20 : 653-667. in the development of the chelonian Bauplan. Fortsch-
ritte der Zoologic, 36 :
ALBERCH P. (1982) - - The generative and regulatory roles
of development in evolution. In "Environmental adap- CANNON H. ( 1 9 5 8 ) - The Evolution of Living Things.
tation and evolution: a theoretical and empirical
Manchester Univ. Press., Manchester.
approach" D. MOSSAKOWSKI & G. ROTH (eds.).
Fischer- Verlag, Stuttgart. CHARLEWORTHE B., LANDER. & SLATKIN (1982) - - A
ALBERCH P. (1983) - - Morphological variation in the neo- Neo-Darwinian commentary on macroevolution. Evolu-
tropical salamander genus Bolitoglossa. Evolution, 37 : tion, 36 : 474-498.
906-919.
CROSSIN K.-L., CHUONG C.-M. & EDELMAN G.-M.
ALBERCH P. (1986) - - Possible dogs. Natural History, 95 : (1985) - - Expression sequences of cell adhesion molecu-
4-9.
les. Proceed. Natn. Acad. Sci. USA, 82 : 6942-6946.
ALBERCH P. & GALE E. (1983) - - Size dependence during
the development of the amphibian foot. Colchicine- CUENOT L. ( 1 9 5 1 ) - L'Evolution biologiqu e. Masson,
induced digital loss and reduction. J. Embryol. exper. Paris.
Morphol., 76 : 177-197.
DOBZHANSKY T. (1951) - - Genetics and the Origin of Spe-
ALBERCH P. & GALE E. i1985) - - A developmental analy- cies, 3rd ed. Columbia Univ. Press, New York.
sis of an evolutionary trend : Digital reduction in amphi-
bians. Evolution, 39 (1) : 8-23. DRACHMAN D.-B. and SOKOLOFF L. (1966) - - The role of
APPLE T. (1987) - - The Cuvier-Geoffroy Debate. Oxford movement in embryonic joint development. Develop.
Univ. Press, London and New York. Biol., 4 : 401-420.
BERG L. (1926) - - Nomogenesis or Evolution Determined DULLEMEIJER P. & BAREL C. (1977) - - Functional mor-
by Law. M.I.T. Press, Cambridge, (1969). phology and evolution. In " M a j o r Patterns in Vertebrate
BERGSON H. (1907) - - Evolution crdatrice. Alcan, Paris. Evolution" M.-K. HECHT, P.-C. GOODY & B.-M.
HECHT (eds.). Plenum Press, New-York.
BLACK S.D. & J.C. GERHART (1986) - - High-frequency
twining of Xenopus laevis embryos from eggs centrifu- EASTON T.-W. (1985) - - A possible mechanism of partial
ged before first cleavage. Devel. Biol., 116 : 228-240. twinning in a calf. Anat. Record, 212 : 100-102.
-- 54 h

EDELMAN G.-M. (1986a) - - Cell adhesion molecules in the tein of neural and non-neural tissues that mediates gila-
regulation of animal form and tissue pattern. Ann. Rev. neuron interaction. Proceed. natn. Acad. Sci. USA, 82 :
Cell. Biol., 2 : 81-116. 8075-8079.

EDELMAN G.-M. (1986b) - - Molecular mechanisms of mor- HALL B.-K. (1978) - - Developmental and Cellular Skeletal
phologic evolution. Chemica Scripta, 26B : 363-375. Biology. Academic Press, New York.
EDELMAN G.-M. (1987) - - Neural Darwinism. Basic Books, HERRING S.-W. & ROWLATT U.-F. (1981) - - A n a t o m y and
New York. embryology in cephalothoracopogus twins. Teratology,
23 : 159-173.
FUTUYAMA D.-J. (1986) - - Evolutionary Biology. Second
Edition. Sinauer Associates, Sunderland, Mass. HINCHLIFFE J.-R. (1977) - - The chondrogenic pattern in
chick limb morphogenesis : a problem of development and
GALLIN W.-J., CHUONG C.-M., F1NKEL L.-H. & EDEL-
evolution. In "Vertebrate limb and somite morphogene-
MAN G.-M. ( 1 9 8 6 ) - Antibodies to L - C A M perturb sis", D. EDE, J.-R. HINCHLIFFE & M. BALLS (eds.).
inductive interactions and alter feather patten and struc- Cambridge Univ. Press, Cambridge.
tures. Proceed. natn. Acad. Sci. USA, 83 : 8235-8239.
HINCHLIFFE J.-R. & JOHNSON D.-R. (1980) - - The Deve-
GARCIA-BELLIDO A. ( 1 9 7 7 ) - H o m o e o t i c and atavistic
lopment o f the Vertebrate Limb. Clarendon Press,
mutations in insects. Amer. Zool., 17 613-629.
Oxford.
GARCIA-BELLIDO A. (1983) - - Comparative anatomy of
cuticular pattern in the genus Drosophila. In " D e v e l o p - HIROMI Y., HUROIWA A. & GEHRING W.-J. (1985) - -
ment and E v o l u t i o n " , B.-C. GOODWIN, N. HOLDER & Control elements of the Drosophila segmentation gene
C.-C. WYLIE (eds.). Cambridge Univ. Press. fushi tarazu. Cell, 3 : 603-613.

GARCIA-BELLIDO A. (1986) - - Genetic analysis of morpho- HO M.-W. & SAUNDERS P.T. ( 1 9 7 9 ) - Beyond neo-
genesis. In "Genetics, Development, and E v o l u t i o n " J.- Darwinism - An epigenetic approach to evolution. J.
P. GUSTAFON, G.-L. STEBBINS & F.-J. AYALA (eds.). theor. Biol., 78 : 573-591.
Plenum Press, New-York.
HSU J. & GONDA M . A . (1980) - - Monozygotic twin for-
GEOFFROY SAINT-HILAIRE I. (1836) - - Trait6 de T6rato- mation in mouse embryos in vitro. Science, 209 : 605-606.
logie. Vols. 1-3. J.-B. Bailli&e, Paris.
INGHAM P.-W. & MARTINEZ-ARIAS A. ( 1 9 8 6 ) - The
GERHART J., BLACK S., SCHARH S., GIMLICH R., VIN- correct activation of Antennapedia and bithorax complex
CENT J.-P., DANILCHIK M., ROWNING B. & ROBERTS genes requires the fushi tarazu gene. Nature, 324 (6097) :
J. (1986) - - A m p h i b i a n early development. Bio-Science, 592-597.
36 : 541-549.
INGHAM P . - W . , ISH-HOROWICZ D. & HOWARD K.-R.
GILBERT S.-F. (1985) - - Developmental Biology. Sinauer (1986) - - Correlative changes in homoeotic and segmen-
Associates, Sunderland, Mass. tation gene expression in Krippel mutant embryos of Dro-
sophila. The EMBO Journal, 5 (7) : 1659-1665.
GOLDSCHMIDT R. (1940) - - The Material Basis of Evolu-
tion. Yale Univ. Press, New Haven. JURILOFF D.-M., SULIK K.-K., RODENICK T . - H . &
HOGAN B.-K. (1985) - - Genetic and developmental stu-
GOODWIN (1982) - - Development and evolution. J. Theo- dies o f a new mouse mutation that produces otocephaly.
retical Biol., 97 : 43-55. J. Craniofacial Genetics Develop. Biology, 5 : 121-145.
GOULD G.-M. & PYLE W.-L. ( 1 9 3 7 ) - Anomalies and KATZ M.-J. ( 1 9 6 3 ) - Ontophyletics : Studying evolution
Curiosities of Medicine. 2 Vols. Sydenham Publ., New-
beyond the genome. Perspectives in Biology and Medi-
York.
cine, 26 (2) : 323-333.
GOULD S.-J. (1977) - - Ontogeny and Phylogeny. Belknap
KATZ M.-J. & LASER R.-J. (1978) - - Evolution of the ner-
Press, Cambridge, Mass., 501 p.
vous system : Role of ontogenetic mechanisms in the evo-
GOULD S.-J. (1983) - - The hardening of the modern synthe- lution of matching populations. Proceed. hath. Acad-. Sci.
e

sis. In "Dimensions of D a r w i n i s m " M. GRENE (ed.). USA, 75 : 1349-1352.


Cambridge Univ. Press, Cambridge.
KATZ M.-J. & LASEK R.-J. & KAISERMAN-ABRAMOF I.-
GOULD S.-J. (1986) - - Archetype and adaptation. Natural R. (1981) - - Ontophyletics of the nervous system : eye-
History, 10 : 16-27. less mutants illustrate how ontogenetic buffer mechanics
channel evolution. Proceed. Natn. Acad. Sci. USA, 78
GRASSt~ P.-P. ( 1 9 7 0 ) - L'6volution du vivant. Albin (1) : 397-401.
Michel, Paris, 477 p.
KAUFFMAN S.A. ( 1 9 7 7 ) - Chemical patterns, compart-
GRUMET M., HOFFMAN S., CROSSIN K.-L. & EDELMAN ments and a binary epigenetic code in Drosophila. Amer.
G.-M. (1985) - - Cytotactin, an extracellular matrix p r o - Zool., 17 : 631-648.
- - 55 m

KAUFFMAN S.A. (1983) - - Internal factors in evolution. In MENDELSOHN E. (1980) - - The continuous and the discrete
" D e v e l o p m e n t and E v o l u t i o n " , B.-C. GOODWIN, N. in the history of science. In " C o n s t a n c y and Chan-
HOLDER & C.-C. WYLIE (eds.) Cambridge Univ. Press. ge in H u m a n D e v e l o p m e n t " , O.-G. BRIM, Jr. & J.
KAGAN (eds.). Harvard Univ. Press. Cambridge, Mass.
KUN1EDA M. & WAKAHARA M. (1987) - - Twin formation
in Xenopus laevis eggs centrifuged before first cleavage. MEYER A. (1987) - - Phenotypic plasticity and heterochrony
Zool. Science, 4 : 489-496. in Cichlasoma managuense (Pisces, Cichlidae) and their
implications for speciation in cichlid fisches. Evolution,
LACK D. ( 1 9 4 7 ) - Darwin's Finches. Cambridge Univ. 1 : 1357-1369.
Press..
MURRAY P . D . F . (1936) - - Bones. Cambridge Univ. Press,
LANCASTER F.-M. (1968) - - Sex-linked winglessness in the
London.
fowl. Heredity, 23 : 257-262.

L A N D E R . (1986) - - The dynamics of peak shifts and the NIEUWKOOP P.D., JOHNEN A.-G. & ALBERS B. (1985) - -
pattern of morphological evolution. Paleobiology, 12 (4) : The Epigenetic Nature of Early Chordate Development.
343-354. Cambridge Univ. Press, London.

LANYON L.-E. & RUBIN C.-T. (1985) - - Functional adap- NIJHOUT H.-F., WRAY G.-A., KREMEN C. & TERAGAWA
tation in skeletal structures. In " F u n c t i o n a l Verebrate C.-K. (1986) - - Ontogeny, phylogeny and evolution of
M o r p h o l o g y " , M. HILDEBRAND, D.-M. BRAMBLE, K- form an algorithmic approach. Syst. Zool., 35 (4) :
F. LIEM & D.-B. WAKE (eds.). Harvard Univ. Press, 445-457.
Cambridge, Mass.
NODEN D.-M. (1986) - - Origins and patterning of craniofa-
LAUDER G. (1981) - - F o r m and function : Structural analy- cial mesenchymal tissues. J. Craniofacial Genetics Deve-
sis in evolutionary morphology. Paleobiology, 7 : 430-442. lop. Biology, Supp. 2 : 15-31.
LENOIR T. (1984) - - The eternal laws of form : Morphoty- NUSSLEIN-VOLHARD C. & WIESCHAUS E. ( 1 9 8 0 ) -
pes and the conditions of existence in Goethe's biologi- Mutations affecting segment number and polarity in Dro-
cal thought. J. Social Biol. Struct., 7 : 317-324. sophila. Nature, 287 : 795-801.
LEWIS E.-B. (1968) - - Genetic control of developmental
ODELL G. & OSTER G., ALBERCH P. & BURNSIDE B.
pathways. Proceed. 12th Internatn. Congress Genetics,
( 1 9 8 1 ) - The mechanical basis of morphogenesis. 1.
1 : 96-97.
A mechanical model of epithelial tissue folding and
certain gastrulation patterns. J. Develop. Biol., 8 5 :
LEWIS E.B. (1978) - - A gene complex controlling segmen-
446-462.
tation in Drosophila. Nature, 276 : 565-570.

LIEM K.-F. (1980) - - Adaptive significance of intra - and OSBORN H. (1934) - - Aristogenesis, the creative principle
interspecific differences in the feeding repertoires of cichlid in the origin of the species. Amer. Nat., 68 : 193-235.
fishes. Amer. Zoo1., 20 : 295-314.
OSTER G. & ALBERCH P. (1982) - - Evolution and bifur-
LOVE JOY A . - O . (1936) - - The Great Chain of Being. H a r - cation of developmental programs. Evolution, 3 6 :
vard Univ. Press, Cambridge, Mass. 444-459.

MACKIE E.-J., THESLEFF I. & CHIQUET-EHRISMANN R. OSTER* G., MURRAY J. & MIANI M. (1985) - - A model for
(1987) - - Tenascin is associated with chondrogenic and chondrogenic condensations in the developing limb. The
osteogenic differentiation in vivo and promotes chondro- role of extracellular matrix and cell fractions. J. Embryo1.
genesis in vitro. J. Cell. Biol., 105 : 2569-2579. Exp. Morphol., 89 : 93-112.
MEINHARDT H. (1982) - - Generation of structures in a OSTER G., SHUBIN N., MURRAY J.-D. & ALBERCH P.
developing organism. In : "Developmental Order : its ori- (1988) - - Evolution and morphogenetic rules : the shape
gin and regulation", S. SUBTELNY & P . - B . GREEN of the vertebrate limb in ontogeny and phylogeny. Evo-
(eds.). A.R. Liss., New York : 439-461. lution, 42 : 862-884.
MEINHARDT H. (1984) - - Models for pattern formation
during development of higher organisms. In "Pattern For- PALLISTER J.-L. (1982) - - Introduction to " O n Monsters
m a t i o n " , G.-M. MALACINSKI & S.-V. BRYANT (eds.). and M a r v e l s " by Ambroise Par& Univ. Chicago Press,
Macmillan Publ. Co., New York. Illinois.

MEINHARDT H. (1986) - - Hierarchical inductions of cell PARTLOW G.-D., BARRALS D.-E. & FISHER K.R.S.
states : A model for segmentation in Drosophila. J. Cell ( 1 9 8 1 ) - Morphology of two-headed piglet. Anat.
Sci. Suppl., 4 : 357-381. Record, 199 : 441-448.
b 56 --

PATOU M.-P. (1973) - - Analyse de la morphogen6se du pied SPURWAY H. (1949) - - Remarks on Vavilov's law of homo-
des Oiseaux /t l'aide de m61anges cellulaires inter- logous variation. In " L a Ricerca Scientifica" (Suppl. Pal-
sp6cifiques. 1. Etude morphologiqne. J. Embryol. Exper. lanza Symposium), n 18, C.N.R. 6dit., Roma.
Morphol., 29 : 175-196.
STEPHENS T.D. & SHEPARD T.-H. (1983) - - A review of
PRAHALD K.-V., SKALA G., JONES D.-C. & BRILES W.E. limb defects in a large fetus collection. Amer. J. Human
(1979) - - Limbless : a new genetic mutant in the chick. Genetics, 35 : 508-519.
J. Exp. Zool., 209 : 427-434,
STERN C. (1954) - - Two of three bristles. Amer. Sci., 42 :
RACHOOTIN S. & THOMSON K.S. (1981) - - Epigenetics, 213-247.
paleontology and evolution. Proceed. Internatn. Cong.
Syst. Evol. Biol., 2, Vancouver (1980). Carnegie-Mellon STOCKARD C.-R. ( 1 9 2 1 ) - D e v e l o p m e n t a l rate and
Univ., Pittsburgh. structural expression : an experimental study of twins,
" d o u b l e monsters" and single deformities, and the inte-
RAFF R.A. & KAUFMAN T.C. (1983) - - Embryos, Genes raction among embryonic organs during their origin and
and Evolution. Macmillan, New York. development. Amer. J. Anat., 28 : 115-266.

RAIKOW R.-J. (1983) - - L o c o m o t o r system. In " F o r m and SUZUKI A.-S., YURIKO M. & KANEDA T. (1984) - - G e r m
Function in Birds", A. KING & J. Mc LELLAND (eds.). layer interactions in pattern formation of amphibian meso-
Vol. 3. Academic Press, New York. derm during primary embryonic induction. Devel. Growth
Differ., 26 : 81-94.
RANZI S. ( 1 9 5 7 ) - Early determination in development
under normal and experimental conditions. In " T h e THIERY J.-D., DELOUVEE A., GRUMET M. & EDELMAN
Beginnings of Embryonic Development", A. TYLER, R.- G.-M. (1985) - - Initial appearance and regional distribu-
C. yon BORSTEL & C.-B. METZ (eds.). Amer. Assoc. tion of the neuron - glia cell adhesion molecule in the chick
Advanc. Sci., 48. embryo. J. Cell. Biology, 100 : 442-456.

RIEDL R. (1979) - - Order in Living Organisms. J. Wiley and TUR1NG A.-M. (1952) - - The chemical basis of morphoge-
Sons, New York. nesis. Phil. Trans. R. Soc., set. B, 237 : 37-72.

RUSSEL E.S. (1916) - - F o r m and Function. John Murray, VAVILOV N.-I. (1922) - - The law of homologous series in
London. variation. J. Genetics, 12 : 47-89.

SAUNDERS P.-T. & HO M.-W. (1976) - - On the increase in VINCENT J.-P., OSTER G. & GERHART J.-C. (1986) - -
complexity in evolution. J. Theor. Biol., 63:375-384. Kinematics o grey crescent formation in the Xenopus egg :
displacement of the subcortical cytoplasm relative to the
SCHINDEWOLF O.-H. ( 1 9 3 6 ) - Pal/iontologie, Entwick- egg surface. Devel. Biol., 114 : 484-500.
lungslehre, und Genetik. Borntrager, Berlin.
WADDINGTON C.-H. (1957) - - The Strategy of the Genes.
SCHINDEWOLF O . - H . (1950) - - Grundfragen der Palfion- Allen and Unwin, London.
tologie. Schweizerbart, Stuttgart.
WAKE D.-B. & LARSON A. ( 1 9 8 7 ) - Multidimensional
SEEGER M.-A. & KAUFMAN T.-C. (1987) - - H o m o e o t i c analysis of an evolving lineage. Science, 238 : 42-47.
genes of the Antennapedia complex (ANT-C) and their
molecular variability in the phylogeny of the Drosophili- WAKE D.-B., ROTH G. & WAKE M . - H . (1983) - - On the
didae. In "Development as an Evolutionary Process", R.- problem of stasis in organismal evolution. J. Theor. Biol.,
A. RAFF & E.-C. RAFF (eds.). Alan Liss. 101 : 211-224.

SHUBIN N.-H. & ALBERCH P. (1986) - - A morphogenetic WATERS N.-F. & BYWATERS J.-H. ( 1 9 4 3 ) - A lethal
approach to the origin and basic organization of the tetra- embryonic wing mutation in the domestic fowl. J. Hered.,
pod limb. Evolutionary Biology, 20. 34 : 213-217.

SHUBIN N.-H. & ALBERCH P. - - The developmental basis WEBSTER G. & GOODWlN B. (1982) - - History and struc-
for skeletal variation in the amphibian limb. Biol. J. Limb. ture in biology. In " T o w a r d s a Liberatory B i o l o g y " , S.
Soc. (in press). ROSE (ed.). Allison and Busby, London.

SLACK J.-M.-W. (1985) - - H o m o e o t i c transformations in WEISS P. (1955) - - Beauty and the beast : life and the rule
man : implications for the mechanism of embryonic deve- of order. Scientific Monthly, December (1955) :286-299.
lopment and for the organization of epithelia. J. Theor.
Biol., 114 : 463-490. WESSELLS N.-K. (1982) - - A catalogue of processes respon-
sible for Metazoan morphogenesis. In " E v o l u t i o n and
SOBER E . & LEWONTIN R.-C. (1982) - - Artifact, cause and D e v e l o p m e n t " , J.-T. BONNER (ed.). Dahlem Konferen-
genic selection. Philosophy of Science, 49 : 157-180. zen (1982). Springer-Verlag, Berlin.
- - 57 - -

WILDER H . - H . (1908) - - The morphology of cosmobia ; WRIGHT S. (1932) - - The roles of mutation, inbreeding,
speculations concerning the significance of certain types crossbreeding, and selection in evolution. Proceed. 11
of monsters. Amer. J. Anat., 8 (4) : 355-440. Internatn. Congr. Genet., 1 : 356-366.
WILLIAMS E.-E. (1972) - - The origin of faunas. Evolution WRIGHT S. (1934) - - Genetics of abnormal growth in the
of Lizard congeners in a complex island fauna : a trial guinea pig. Cold Spring Harbor Symp. on Quantitative
analysis. Evolutionary Biology, 6 : 47-89. Biology, 2 : 137-147.
WILLIAMS E.-E. (1983) - - Ecomorphs, faunas, island size
and diverse end points in island radiations of Anolis. In WRIGHT S. & WAGNER K. (1925) - - Types of subnormal
" L i z a r d Ecology. Studies of a Model O r g a n i s m " , R.-B. development of the head from inbred strains of guinea
HUEY, E.-R. PIANKA & T.-W. SCHOENER (eds.) Har- pigs and their bearing on the classification and interpre-
vard Univ. Press, Cambridge. tation of vertebrate monsters. Amer. J. Anat., 54 (3) :
383-447.
WIMSATT W.-C. (1980) - - Reductionist research strategies
and their biases in the units of selection controversy. In ZWILLING E. (1955) - - Teratogenesis. In "Analysis of Deve-
"Scientific Discovery : case Studies", T. NICKLES (ed.). l o p m e n t " , B.-H. WlLLIER, P.-A. WEISS & V. HBUR-
D. Reidel Publ. GER (eds.). W.-B. Saunders, Philadelphia : 699-719.

S-ar putea să vă placă și