Evolution, 36(3), 1982, pp.

427-443

CHARACTER CHANGE, SPECIATION, AND THE HIGHER TAXA

SEWALL WRIGHT
Laboratory of Genetics, University of Wisconsin, Madison, Wisconsin 53706

Received December 9, 1980

Gould and Eldredge (1977) in their ar- chical process with complementary but
ticle, "Punctuated Equilibria: The Tempo different modes of change of its three ma-
and Mode of Evolution Reconsidered," jor levels, variation within populations,
took issue with the common view of phy- speciation and patterns of macroevolu-
letic gradualism. They agreed to a consid- tion." .
erable extent with the view of Simpson, Simpson definitely rejected Gold-
to whose book, published 36 years ago, schmidt's (1940) thesis that speciation and
their title referred. Simpson (1944) brought the origin of the higher categories depend
out the enormous differences among rates on types of mutation that have nothing in
of evolution indicated by paleontological common with the changes that occur
data: the near stasis (bradyte1y) of some within species.
forms for hundreds of millions of years, I am not in a position to discuss inde-
typical rates (horotely) (which, however, pendently the data of paleontology and
vary much among phyla) and the enor- recognize that my field, genetics, bears
mously rapid rates (tachytely) indicated directly only on microevolution, but I feel
especially for the periods of origin of many that we should explain phenomena at the
important groups. He used the term higher levels as far as possible, as flowing
"quantum evolution" for evolutionary from observed phenomena of genetics in
events of the sort referred to as "punctua- the broad sense, including cytogenetics,
tional" by Gould and Eldredge. The latter before postulating wholly unknown pro-
began the abstract of their paper with the cesses. This does not bar me from accept-
statement: "We believe that punctuational ing selection among entities at allleve1s of
change dominates the history of life: evo- the biological hierarchy.
lution is concentrated in very rapid events
of speciation (geologically instantaneous Historical Review
even if tolerably continuous in ecological The opposed concepts, gradual and
time." abrupt change, go back to the origins of
This does not contrast as much as it evolutionary thought. At the beginning of
may seem at first sight with Simpson's as- the last century, Lamarck postulated a
sertion (p. 203) that "nine tenths of the preordained ladder of life, leading from
pertinent data of paleontology fall into the simplest forms to man with gradually
patterns of the phyletic mode," since he diverging branches, determined by the in-
held that there might be episodes of tachy- heritance of characters acquired by ad-
tely in the phyletic evolution. There is aptation to different conditions. E. Geof-
more contrast, however, with respect to froy St. Hilaire, on the contrary, proposed
mode. Gould (1977) in his paper, "The that radically different patterns of life had
Return of Hopeful Monsters," wrote: "I appeared abruptly from time to time, a
do predict that during the next decade, view derived from the observed appear-
Goldschmidt (1940) will be largely vindi- ance of monstrosities. These speculations
cated in the new world of evolutionary came when little was known of the actual
biology." course of evolution (d. Nordenskiold,
In his paper, "Is a New and General 1928).
Theory of Evolution Emerging?" (1980), Darwin (1859), in mid nineteenth cen-
he maintains that "Evolution is a hierar- tury, was able to marshall the available

427
428 SEWALL WRIGHT

data from all fields of biology in such a marckiana, that had escaped from culti-
way as to win almost immediate accep- vation in the Netherlands. It turned out
tance by scientists of evolution as a fact. later, however, that most of his supposed
He also presented a theory that was at new species were trisomies with merely
least in the spirit of physical science. He altered proportions of the elements of he-
held that study of the ways in which an- redity of O. Lamarckiana, not transmis-
imal and plant breeders had actually sible, moreover, by pollen, so that the new
brought about striking changes by artifi- "species" could exist only as segregants.
cial selection provided insight into the One mutant form, O. gigas, turned out,
evolutionary process in nature. After care- however, to be a tetraploid, capable of
ful consideration (Darwin, 1868 Vol. 2) he reproducing itself and producing sterile
concluded that: "Without variability, triploid hybrids on backcrossing. It thus
nothing can be effected: slight individual really did behave like a new reproduc-
differences, however, suffice for the work tively isolated species, although differing
and are probably the chief or sole means morphologically from the parent species
in the production of new species." only very slightly. There was thus evi-
Thus Darwin came down strongly on dence that a new species might arise
the side of gradualness under natural se- abruptly.
lection though he did not wholly rule out The rediscovery of Mendelian heredity
an occasional role of the more striking in 1900 soon dissipated the first of the ob-
changes that he called "sports." jections to Darwin's theory, referred to
The adequacy of Darwin's explanation above. Yule pointed out in 1902 that seg-
of evolution was widely questioned, prin- regation in the ratio 1:2:1 in F 2 of a cross
cipally for two reasons: (1) the rapid di- persisted in randombred F 3 and in later
lution of variation, under the prevailing randombred generations. Castle (1903) ex-
theory of a blending of the parental he- tended this to ratios based on other gene
redities, seemed to require an unbelieva- frequencies and Weinberg (1908) and
ble efficiency of the selection process, an Hardy (1908) put this in general mathe-
objection put in mathematical form by matical form. Under Mendelian heredity
Fleeming Jenkin (1867), and (2) the diffi- there is no such dilution of hereditary ele-
culty of accounting for the extraordinary ments as implied by the theory of blending
coadaptiveness of all parts of organisms, heredity.
urged especially by St. George Mivart. The study of conspicuous Mendelian
Many biologists continued to accept a difference, led most of the early geneticists
perfecting principle as an essential prop- to accept an attenuated form of the mu-
erty of life (e.g. Nageli), a view fervantly tation theory: it was supposed that gene
advocated as late as 1934 by the leading mutations with major effects were occa-
paleontologist at the time, H. F. Osborn, sionally favorable and that these replaced
in his doctrine of aristogenesis. the old type genes, one at a time. This was
The majority, however, followed the a form of 'punctuated' evolution but much
paleotologist Cope in preferring the inher- less drastic than that of de Vries.
itance of the effects of use and disuse (d. A few continued to accept speciation by
Kellogg, 1907). mutation and some, notably Goldschmidt
At the turn of the century, de Vries pro- (1940), went beyond de Vries in holding
posed his mutation theory under which that the origin of higher categories re-
species appear abruptly, natural selection quired mutations of appropriate sorts.
being relegated to the role of guiding the Castle, with an agricultural back-
course of evolution beyond the species ground, differed from the prevailing Men-
level. de Vries's theory had the merit of delian view in the opposite direction. He
being based on the actual appearance of challenged the current belief, tracing to de
what seemed new species among plants of Vries, according to which the selection of
the American species, Oenothera La- quantitative variability could produce no
 MACROEVOLUTION-SmFTING BALANCE THEORY 429

permanent effects. He carried through that are the primary products of DNA ac-
large scale experiments with a strain of tivity, may have no functional significance
hooded rats in which he attempted to beyond filling space and that this may ac-
change the pattern of black and white by count for the extensive polymorphism of
selection, experiments in which I was his proteins revealed by electrophoresis in
assistant during 1912-1915. He was ap- many species. This was obviously, how-
proaching self-black in 20 generations of ever, not a general theory of gene-char-
plus selection, self-white in the same num- acter relations.
ber of generations of minus selection, when Figure lA represents the one-to-one re-
reduced fecundity brought both lines to an lation of gene and "unit character" pos-
end (Castle, 1916). My prior acceptance tulated in the prevailing Mendelian view
of Darwin's views on the efficacy of selec- of evolution, referred to above. The cours-
tion of quantitative variability was con- es taken under constant selection, but di-
firmed though I was somewhat disturbed verse assumptions otherwise, were worked
by the forced termination of the experi- out systematically by Haldane in a series
ments in spite of much heritable variabil- of papers, beginning in 1924 and sum-
ity, which suggested that selection tended marized in 1932, following the working
to have deleterious pleiotropic conse- out of special cases by Castle (1903) and
quences. H. T. J. Norton (1915).
Such a one-to-one relation could, how-
Gene-Character Relations eyer, only be assumed for genes with ma-
Castle originally thought that the quan- jor effects. Haldane also worked out con-
titative variability was in the piebald fac- sequences of other assumptions, including
tor itself, a view confirmed to some extent multifactorial heredity. Figure IB repre-
by the appearance of a new allele by mu- sents the latter hypothesis, uncomplicated
tation in the plus series, an allele with by gene interaction and pleiotropy. Fisher
effects intermediate between that of the (1930) presented convincing reasons for
allele in this series and that of the self- holding that only minor factors are likely
colored wild rats. Meanwhile, however, to be favorably selected. He accepted the
the multiple factor theory of quantitative occurrence of nonadditive relations be-
variation, foreshadowed by Mendel him- tween gene and character (dominance, ep-
self and proposed by Yule in 1906, had istacy) but showed that these merely slow
been exhaustively demonstrated in a case down progress by selection under his as-
in wheat by Nilsson-Ehle (1909) and more sumption that species are effectively
extensively in maize and Nicotiana by panmictic. Fisher, in contrast with Hal-
East (1910, 1916) and his associates. This dane, attempted to bring all evolution un-
provided an alternative interpretation of der a single simple formula, his "funda-
Castle's (1919) results which he accepted mental theorem of selection," "the rate of
after carrying through an extensive test. increase in fitness of any organism at any
The Mendelian interpretation of natural time is equal to its (additive) genetic vari-
selection requires analyses of the statistical ances in fitness at that time."
consequences of diverse assumptions on It is not wholly fair to assert that Fish-
the relations of genes to characters. Figure er's conception of the relation of genes to
1 (Wright, 1980) gives a diagrammatic il- characters was restricted to that of Figure
lustration of a number of different as- IB but he clearly tended to think in those
sumptions. Figure l C applies to cases in terms. Thus his assumption (1929) that
which a block of DNA (left) does not code "a small selection of intensity, say 1/50,000
for anything, or anything of significance, the magnitude of a larger one, will pro-
and thus has no phenotypic effect (right) duce the same effect in 50,000 times the
that is subject to selection. Kimura (1968) time," while true of gene character-rela-
and King and Jukes (1969) suggested that tions of the sort represented in Figure IB
large portions of the protein molecules (and lA) is not at all true of genes that
430 SEWALL WRIGHT

modify a heterozygote and also have pleio- On taking account of the pleiotropic ef-
tropic effects on the homozygotes. Fisher fects on eumelanin and phaeomelanin, and
(1928) made the above assumption in con- the likelihood of other slight morphologi-
nection with his theory that the prevailing cal and physiological effects, there can be
dominance of wild type over recurrent little doubt that pleiotropy is universal.
deleterious mutations is due to specific The situation is similar in other organ-
modifiers of the rare heterozygotes, exert- isms.
ing a selective pressure of the order of the My own speculations on evolution were
mutation rate, put at 10-6 per generation. dominated from the first by the thought
His principal evidence for his theory was that there must somehow be selection of
the easy modifiability of intermediate het- coadaptive interaction systems as wholes.
erozygotes under strong direct selection. The difficulty was that under biparental
The latter well known phenomenon is ob- heredity the reduction division breaks up
viously irrelevant if there are always at combinations so rapidly in terms of geo-
least slight pleiotropic effects on homo- logic time, that under panmixia natural
zygous wild type (Wright, 1929a, 1977; selection is capable of operating only on
Charlesworth, 1979). the average effects of genes in all combi-
Figure ID represents each gene as hav- nations. Combinations of unlinked genes
ing multiple pleiotropic effects because of go halfway toward randomness per gen-
interaction of its products with those of eration; with 10% recombination, half way
others. It may be assumed, moreover, that per 7 generations; and with 1% recombi-
the effects of these interactions are not in nation half way per 69 generations.
general additive. These are inevitable
consequences of the complex network of The Selective Topography
biochemical and developmental reactions The nature of the field of variability
that intervene between primary gene ac- available for natural selection under the
tion and the ultimate effects subject to se- various patterns of gene-character rela-
lection. tions requires consideration.
The available evidence is in harmony Under the concept of one-to-one rela-
with. this concept. My own major experi- tionship (Fig. lA), it should be possible
mental project from the time when I was theoretically to rank the alleles at each lo-
a graduate student (cf. Wright, 1968), has cus in the order of their values to the or-
been the study of gene interactions in ganism. Natural selection would operate
guinea pigs. Thus I have made many under given conditions according to the
thousand different combinations of the courses described most systematically by
genes at 11 loci that affect coat and eye Haldane as already noted.
color and considerable numbers of loci af- An organism, however, is very far from
fecting other characters. From the first, I being a mosaic of unit characters. The
was fascinated by the seeming unpredict- value of any gene depends in general on
ability of such combinations and have the array of other genes with which it is
tried to devise hypothetical interaction associated. This holds even under the pat-
patterns to account for them. There were tern of relations of Figure IB in which the
also often surprising pleiotropic effects. effects of multiple loci are additive with
The latter have been even more striking respect to characters but usually there is
in the studies of the genetics of the mouse an intermediate optimum close to the
by many workers, in which enormously character mean.
more new mutations, usually deleterious, Assume a group of genes A, B, C and
have been observed. More than 50 loci D that contribute equally and additively
have been found that affect coat color, of to the size of some part, relative to their
which some 60% also have gross morpho- alleles, a, b, c and d (Fig. 2) (Wright,
logical effects (Searle, 1968; Silvers, 1979). 1964b). Note that only the positive factors
 MACROEVOLUTION-SHIFTING BALANCE THEORY 431

SPECIAL CASES GENERAL W

1.25 f-

1.00 -

0.75 -
A. ONE TO ONE; MAJOR B. POLYGENIC VARIATION
HALDANE (1924, 32) FISHER (1926, 30)

0.50-

0.25~

0
C. NEUTRAL AMINO ACIDS D.NETWORK A ABC
KIMURA (1968) WRIGHT (1929,31,32) B AB,BC ABD
C AC,BD ACD
0 D AD,CD BCD ABCD
FIG. 1. Four assumptions on the relationship of
smallest optimum largest
genotype (left) to phenotype (right).
FIG. 2. Contributions to selective value of the
combinations of four pairs of alleles to a quantita-
tively varying character, assuming equality and ad-
are shown and these singly in the homo- ditivity with optimum in the middle. AB represents
zygotes. There is obviously no best allele AABBccdd, etc. (from Wright, 1964b, Fig. 8).
at any locus. Fixation of any two positive
genes and of the negative ones at the other
two loci gives the mean and hence the op- AC, BD and AD) and one lowest maxi-
timum. There are six such optimal com- mum, CD. The four homallelic combina-
binations (called selective peaks later): tions with only one positive gene and the
AABBccdd, AAbbCCdd, AAbbccDD, four with three positive genes are still low-
aaBBCCdd, aaBBccDD and aabbCCDD. er. That with all four positive genes is
The mean is also given by many other much lower and that with all four nega-
combinations such as AaBbCcDd but these tive genes is the lowest of all.
involve heterozygosis and would give some Imagine now a figure with four orthog-
inferior offspring. enal dimensions, one for each indepen-
In this case, it would make no differ- dent gene frequency and imagine a fifth
ence which one of the six optimal types dimension for selective values. Figure 4
becomes established by selection. In an un- shows two of the faces, including the low-
fixed population, selection (according to est maximum CD, one of the four inter-
Fisher's fundamental theory) will fix the mediate ones, BC and the highest one,
combination to which the composition of AB.
the population is closest. The values chosen for the pleiotropic
In actual cases, however, the maxima effects determine a saddle between peaks
would not be equally fit, whether because CD and BC and one between BC and AB.
of unequal gene effects or because of dif- Arrows indicate the trajectories of popu-
ferent pleiotropic effects. Let us assume lations subject to the assumed selection
that the effects are equal but that A and pressures. If it is assumed that there is
B have certain pleiotropic effects that are recurrent mutation or a small amount of
equal and additive. Figure 3 shows the immigration from other demes, this pre-
selective values of the 16 homallelic pop- vents permanent fixation at any peak.
ulations on a vertical scale. With the cho- In Figure 5 the selection values of Fig-
sen pleiotropic effect, there is one fittest ure 4 are shown according to a vertical
type AB, four at the same lower level (B C, scale. Selective values are also shown along
432 SEWALL WRIGHT

AB ABC AB
1.000
.~.-. ....
1.250

+~'\ -;/1' +
~~ ; i t 1
BCD 1.115 , , '
0.875
e-.. ~
H:~
....... ...--
t
.
~~ - ,4/ t 1.125 0.875
T
~\
-
}f
//.
tt
1 Be. B

...
0.990 \.\
"'.-.. . . .....-. _ t
1.000
~.-.

0.750
CD C

FIG. 4. Two of the surfaces of the 4-dimensional

space of frequencies of genes A, B, C and D of Fig.
3. Trajectories are indicated by arrows. The selective
values at the peaks, pits and saddles are given (from
Wright, 1964b, Fig. 10).
FIG. 3. Total selective values of the 16 homal-
lelic genotypes of Fig. 2 according to the contribu-
tions indicated there, supplemented by equal semi-
dominant pleiotropic effects of genes A and B (from In my 1932 paper, the first in which the
Wright, 19Mb, Fig. 9). concept of a selective topography was pre-
sented graphically, it was stated that "The
problem of evolution, as I see it, is that of
paths from one peak to a higher one, pass- a mechanism by which the species may
ing through the saddle: A path directly continually find its way from lower to
from lowest to highest peak passes across higher peaks in such a field. In order that
the 4-dimensional saddle, but this is more this may occur, there must be some trial-
depressed and thus presents a greater ob- and-error mechanism on a grand scale by
stacle to a peak-shift than do the paths which the species may explore the region
passing through the 2-dimensional sad- surrounding the small portion of the field
dles. which it occupies."
Since quantitative variability with op-
timum close to the mean is the usual rule Modes of Evolution
for measurable characters in wild species, At this point I will go back to a group
the foregoing model is practically univer- of six diagrams (Fig. 7) that I used in 1932,
sal for such characters. It has to do, how- and often later, for consideration of pos-
ever, with only small changes. sible modes of evolution under Mendelian
Of greater importance for major evo- heredity. In these diagrams the multiple
lutionary changes are probably the unpre- dimensions of gene frequency change,
dictable major interaction effects of genes, provided by thousands of heterallelic loci,
referred to earlier, for which Figure ID are flattened out into two dimensions on
with the full complexity expected from the which all genotypes are supposed to be
network of biochemical and develop- located and the dimension of selective val-
mental processes, is intended to be the ue is represented by contours. The innu-
model. The selective topography would be merable selective peaks are given token
correspondingly more complex. representation in only two.
Of special importance are interaction The three upper diagrams represent
systems consisting of a recurrent major cases in which the number of individuals
mutation and one or more modifiers that in an effectively random breeding popu-
neutralize its inevitable deleterious side lation is so great that accidents of sam-
effects. A shallow saddle may lead to a pling can play no significant role. In Fig-
great step in advance (Fig. 6). ure 7A it is assumed that recurrent
 MACROEVOLUTION-SlllFTING BALANCE THEORY 433

-
r andom
drift
random
drift
--
r andom
ultimate
selective peak
drift initial MABCDE
selective peak
\-···1····_···+..
selection
·+····_··--\
selection
MABCD
+ A AB ABC
ABCD ABCDE
AB MABC
w
MAB

1 MA

FIG. 5. Profile of trajectories from peak to peak M

(CD to BC to AB) through 2-dimensional saddles . FIG. 6. Comhinations of an initially deleterious,
and from CD to AB through a 4-dimensional saddle but potentially favorable, major mutation, with fa-
and from pits to peaks (straight lines) (from Wright, vorable modifiers (from Wright, 1964b, Fig. 13).
1964b, Fig. 11).

which it has reached for historic reasons,

mutation is balanced by feeble selection. comes to occupy a region of the field about
The population occupies a small portion this peak and stays there without further
of the field (indicated by a broken circle) change except in so far as novel favorable
about the joint equilibrium point. An in- mutations create a change in the topog-
crease in mutation rate would bring about raphy, an exceedingly slow process. With
an increase in the portion of the field oc- increased selection, the region occupied
cupied (indicated by a solid line). There decreases.
is no appreciable chance, however, that If, however, conditions change quali-
this will come under control of a higher tatively (Fig. 7C), the topography changes.
peak. A decrease in selection, such as oc- The species tends to move with movement of
curs when a particular character ceases to the peak that it has occupied. As I noted
be of value, has a similar effect. The char- in 1932: "Here we undoubtedly have an
acter may wholly disappear under the important evolutionary process and one
mutation pressure but selection for pleio- which has been generally recognized. It
tropic effects of the genes is likely to be consists largely of change without advance
much more important beyond the point of in adaptation. The mechanism is, how-
selective advantage from degeneration of ever, one which shuffles the species about
the character as an encumbrance (Wright, in the general field. Since species will be
1929a, 1964a). Mutation pressure by itself shuffled out of low peaks more easily than
is probably of little importance for char- high ones, it should gradually find its way
acters though it probably is for useless to the higher general regions of the field
genes. as a whole."
Figure 7B represents the case of mass The process of change without advance
selection in an effectively panmictic pop- is what Van Valen has called "the Red
ulation, living for a long time under the Queen Process." The process as a whole
same conditions. The population moves is that under which selection according to
toward the selective peak, the slope of Fisher's fundamental theorem is most ef-
 434 SEWALL WRIGHT

/e"",\ "-
(",... //.,;:::::? (.·.(s.·~.~.·.·.~.··,.v-",\~.:: .
- . .

·~~~~;;I r((~~
'.
....•............
. ~
: ..
./'
"
..
.......................) :/·:(i: .i.:.±~.:.·.~j·
./ ,/.. .
A.lncreased Mutation B.Increased Selection C. Qualitative Change
or reduced selection or reduced mutation of environment
4Nu, 4Ns very large 4 Nu, 4 Ns very large 4 Nu, 4Ns very large

-. .. . \ ... -::
.....................

./~':::=
.........................~
../.i
-----;:} ((({{W
D.Close Inbreeding E.Slight Inbreeding F. Division into local Races
4Nu, 4Ns very small 4Nu,4Ns medium 4nm medium
FIG. 7. Token representation of a portion of the multidimensional array of genotypes of a population
with fitness contours. Field initially occupied indicated by heavy broken contour. Field occupied later
indicated by crosshatched area (multiple subpopulations in F). Courses indicated in C, D, E and F by
arrows. Effective population numbers, N (total), n (local); v (mutation), s (selection), m (migration) (from
Wright, 1932, Fig. 4).

fective (more effective than under the un- followed by many others (including text-
changing conditions of Fig. 7B). It is the books published in 1979 and 1980), have
process which such recent authors as attributed to me the view that fixation of
Maynard Smith (1975), Williams (1966) nonadaptive characters by random drift
and Dawkins (1976) consider all-impor- was the essence of my theory.
tant. It should be noted that I was consider-
Figure 7D refers to populations that ing the evolution of ordinary characters,
have become so small that accidents of not completely neutral primary gene ef-
sampling overwhelm all but the strongest fects such as Kimura (1968) and King and
selective differences. The population wan- Jukes (1969) have discussed in recent years.
ders from the selective peak that it has Figure 7E represents the case of loci
occupied, moves about irregularly, de- with respect to which the effects of acci-
creases in variability. As I noted in 1929, dents of sampling and of selection are
(as essentially in 1931 and later): "In too about equal within a rather small isolated
small a population, there is nearly com- population. I noted of this in 1932: "The
plete random fixation, little variation, lit- species moves down from the extreme peak
tle effect of selection and thus a static con- but continually wanders in the vicinity.
dition, modified occasionally by chance There is some chance that it may encoun-
fixation of a new mutation, leading to de- ter a gradient leading to another peak,
generation and extinction." shift its allegiance to this. Since it will es-
In spite of this, most authors including cape relatively easily from low peaks-
Huxley (1942) and Fisher and Ford (1947) there is here a trial-and-error mechanism
 MACROEVOLUTION-SHIFTING BALANCE THEORY 435

by which in time the species may work its

.,
CD
way to the highest peak in the general ~AB "" 1.00
- 0 -I
field. The rate of progress is extremely
slow, however, since change of gene fre- ~ ,,
" .; 1.25.....
BC
"
.-
\
quency is of the order of the reciprocal of o '- AD "
the effective population size and this re- 1.125 ~ ~ / ;
o
/ 1.125 ......
ciprocal must be of the order of the mu- , 'o~AB' 1 It"
tation rate in order to meet the conditions ; - 1.25' ,
t ; " ,'-AC
of this case." .# 0 CD
1.125 1.00
Finally, Figure 7F represents a species
I .\ "
that is subdivided into local populations
(demes), sufficiently small and isolated that
accidents of sampling may overwhelm FIG. 8. Diagram of a population range, charac-
terized initially by the lowest peak, CD of Fig. 4 in
many weak selection pressures. There which intermediate peaks, BC, AC and AD have
must, however, be enough diffusion that been arrived at locally and the highest peak AB has
a deme that happens to acquire a favor- been arrived at from BC and from overlap of BC
able interaction system may transform its and AC (from Wright, 1964, Fig. 12).
neighbors to the point of autonomous es-
tablishment of the same peak, and thus
ultimately transform the whole species or curred and are spreading by selective dif-
at least that portion of it in which the new fusion, according to their times of origin.
system actually is favorable. The field of The highest peak, AB, has been arrived
variability of the species is here amplified at both within that characterization by BC
by local differentiation, and natural selec- and by overlapping of the regions con-
tion is amplified by the selective diffusion trolled by BC and AC.
from the superior demes. It should be added that I suggested in
The process within each of the demes 1931 and demonstrated later (Wright,
is somewhat similar to that in Figure 7E, 1948) that random differentiation of local
but is limited here only by the immigration populations may be due to fluctuations in
rate which may be thousands of times the the systematic pressures of selection and
mutation rate. If there are thousands of of amount and kind of immigration with-
sufficiently independent demes, the pro- out there being small size populations.
cess in the species as a whole may be mil- Random drift from accidents of sampling,
lions of times as effective as in Figure 7E. however, affects all of the thousands of
This is the process that I later called the loci not subject to strong selection while
shifting balance process. fluctuations in selection would affect a
I have devoted many papers (with con- more limited number.
clusions summarized in Wright, 1969
Chapter 12) to the conditions under di- Selection by Man in Relation to
verse population structures (continuous, Evolutionary Theory
clustered, 'island,' multiple colonies sub- Darwin in "The Variation of Animals
ject to frequent extinction and refounding and Plants under Domestication" or-
from the superior ones) under which the ganized his discussion of "selection by
process may be effective. man" under three heads: methodical, un-
Going back to the model with factors conscious and natural, all concerned with
A, B, C and D, the operation of the shift- direct selection among individuals. He did
ing balance process within the range of the not discuss the sort of selection practiced
species is represented in Figure 8. It is as- by breeders in purchasing breeding stock
sumed that the species has been under from superior strains. The only references
control of the lowest of the six selective to this are in quotations. Thus in a section
peaks (CD). Shifts to control by certain on "selection in ancient and semicivilized
higher peaks, BC, AC and AD have oc- people" he quotes Virgil "as giving as
436 SEWALL WRIGHT

strong advice as any modern agriculturist Shorthorn breed of cattle by means of a

could do, carefully to select the breeding previously devised inbreeding coefficient
stock 'to note the tribe, the lineage, and (Wright, 1923; McPhee and Wright, 1925)
the sire, whom to reserve for husband of led directly in 1925 to the two level shift-
the herd. '" He also quotes verses from ing balance theory of evolution in nature,
ninth century Ireland describing a ransom though publication was delayed in 1931
demanded by Cormac: (abstract 1929). The general conclusions
were somewhat revised in 1932 on the ba-
Two pigs of the pigs of MacLir
sis of the concept of a selective topography
A ram and ewe both round and red,
along the line brought out in the preceding
I brought with me from Aengus,
section. The mathematical treatment in
I brought with me a stallion, and a mare
the 1931 paper dealt only with certain as-
From the beautiful stud of Manannan,
pects of the simplest model. This was re-
A bull and a white cow from Druim
medied and conclusions on various special
Cain.
aspects arrived at in papers in the next
There is clear recognition of selective two decades but the general conclusions
diffusion from superior herds in a quota- still held (cf. Wright, 1977).
tion from a Mr. Wilson to the effect that These conclusions had to do with char-
in certain districts of the Scottish High- acter change within species (microevolu-
lands: "The breeding of bulls is confined tion). I had, however, accepted from the
to a very limited number of persons, who first the concept of speciation as repro-
by devoting their whole attention to this ductive isolation. I suggested in papers in
department are able from year to year to 1940 (abstract 1938) and 1941a that incip-
furnish a class of bulls which are steadily ient speciation from chromosomal rear-
improving the general breed of the dis- rangement depends on an extreme pattern
trict." Darwin presented this quotation to of population structure that is also very
illustrate "the great principle of division favorable for character changes by means
of labor" rather than that of selective dif- of peak-shifts.
fusion from superior herds. I tacitly treated macroevolution as
I do not mean to imply that Virgil, the merely an extension of evolution within
ancient Irish author and Mr. Wilson rec- species in my early papers but in reviews
ognized an important aspect of selection of books by Willis (Wright, 1941b), Gold-
overlooked by Darwin. He undoubtedly schmidt (Wright, 1941c) and Simpson
recognized this aspect but the grading up (Wright, 1945), all concerned especially
of inferior stock by sires drawn from su- with macroevolution, I held that the de-
perior herds had no such clear analogy to termining factor was in general an unusu-
anything happening in nature as did se- al ecological opportunity, not any sort of
lection among individuals. It was more- unusual mutation, an opportunity consist-
over, impossible to grasp fully the quali- ing of the occurrence for one reason or
tative difference between the merely genic other, of vacant ecological niches, more or
consequences of selection among individ- less related to that occupied by the species
uals and the organismic consequences of in question. The occupation of these in a
selection among differentiated herds, be- rapid adaptive radiation constituted the
fore the mechanism of heredity had come origin of a new higher taxon. These con-
to be understood. This is still not grasped cepts will be elaborated in the following
by those who urge strongly the importance sections.
of selection for coadaptation but reject the
only process, interdeme selection, by which Speciation
this can occur in nature under biparental Discussion of speciation had best begin
reproduction. with the recognition that the term "species"
Studies of the breeding history of the has come to mean something very differ-
 MACROEVOLUTION-SIDFTING BALANCE THEORY 437

ent, at least in principle, to neontologists leading to more or less aneuploidy and

from what it is possible for it to mean in thus incipient separation; and (4) there
practice to paleontologists. Before the be- may be nucleo-cytoplasmic incompatibil-
ginning of the present century, there was ity which I will not discuss further.
no such divergence. The species were the Under the first head, mating may be
kinds of organisms, distinguishable by prevented by geographic isolation, by dif-
clear-cut differences in morphology (ex- ference in the breeding seasons, by differ-
cluding "varieties" known to occur within ence in ethological patterns such as court-
progenies). ship, by ecological difference within the
It came to be recognized by such sys- same region 'or by anatomical incompat-
tematists as Osgood (1909) that reproduc- ability. Finally there may be biochemical
tive isolation and thus incipient evolution- bars to fertilization.
ary branching, constituted a criterion of Under the second head, faulty devel-
more significance biologically. A species opment or low fecundity may occur in F 1
came to refer to a population or group of but if F 1 is normal, the irregular propor-
intergrading populations, reproductively tions of the components of the two hered-
isolated from all other such populations. ities in F 2 and later may lead to enough
Intergrading populations might differ faulty development or low fecundity to
greatly at their centers, but should be des- bring about at least incipient speciation.
ignated as merely subspecies, while ones Under both of these heads, the differ-
that could be demonstrated to be repro- ence may be due either to a single locus
ductively isolated, even though showing or may require passage of a threshold after
no consistent morphological difference, accumulation of slight effects at many
should be considered separate species. loci. There is no essential difference from
There are, of course, many intermediate microevolutionary changes except in that
cases, difficult to classify, but the identi- there is some contribution to reproductive
fication of the branching points of the evo- isolation.
lutionary process as accurately as possible Chromosomal differences are conve-
came to be considered the primary objec- niently divided into balanced and unbal-
tive. anced. The latter includes duplications
This concept necessarily breaks down and losses either of whole chromosomes or
in the attempt to name successive species of portions down to single loci. The un-
in a chain along which the morphological likelihood of speciation involving whole
differences ultimately become so great that chromosomes has already been touched on
recognition of speciation becomes imper- in connection with de Vries' Oenothera
ative. If change has occurred at a uniform mutations.
rate, the boundaries between species are Deficiencies of any extent are usually
necessarily arbitrary. If, on the other hand, lethal when homozygous and thus also
there have been periods of apparent stasis, have little or no significance for evolution.
separated by apparently abrupt, or at least Small duplications on the other hand, are
very rapid, change, the boundaries are usually viable even as homozygotes. Typ-
made most conveniently at the latter times. ically they have effects similar to those of
Reproductive isolation may be brought gene mutations and are transmitted as if
about in several different ways (d. Dob- Mendelian genes. Many of them probably
zhansky, 1970):(1) union of egg and sperm become inactivated ultimately. Those that
from the two populations may be pre- persist tend to become differentiated and
vented in various ways; (2) if fertilization increase the stock of loci. They are un-
occurs, normal development may fail be- doubtedly important in evolution but not
cause of imperfect cooperation of the two with respect to speciation.
heredities; (3) chromosome rearrange- This brings us to the balanced chro-
ments may interfere with normal meiosis, mosome changes. Here, as noted earlier,
438 SEWALL WRIGHT

we encounter a real abrupt species-form- majority of these cases, the structural

ing mechanism, the duplication of the en- chromosomal rearrangements have played
tire genome. If occurring in a hybrid be- a primary role in initiating divergence."
tween species whose chromosomes have On the other hand, there are many cases
become so differentiated that none of them in which a chain of intergrading subspe-
pair in meiosis, amphidiploids are formed, cies has returned on itself in a circle and
capable of breeding true to the hybrid it is found that the overlapping popula-
phenotype and reproductively isolated tions coexist without interbreeding as if
from both parent species because of the distinct species (Osgood, 1909; Mayr,
sterility of the triploids resulting from 1963). Extinction of the intermediates
backcrosses. There is no doubt that a would elevate them to this rank without
great many species of plants, though few any change whatever in themselves. These
of animals, had this origin. On the other cases indicate that not all speciation de-
hand, the autotetraploids formed from pends on chromosomal rearrangement.
doubling the genome of a single species Returning to the latter, there is a prob-
differ little from the parent in phenotype. lem as to how a new chromosomal rear-
They can reproduce their type and pro- rangement can pass the barrier imposed
duce sterile triploids in backcrosses but by strong selection against the hetero-
the sets of four chromosomes cause irreg- karyons. White (1978) and Hedrick (1981)
ularities in meiosis. They are, no doubt, lean to meiotic drive from asymmetrical
responsible for many new species but not segregation, as the explanation.
as many as amphidiploids. Allotetraploids Fixation by accidents of sampling is,
with pairing of some but not all chromo- however, another possibility. It can occur
somes resemble full amphidiploids in orig- only in extremely small colonies (Wright,
inating a morphologically new species but 1940, 1941a). The most favorable situa-
resemble autotetraploids in having some tion is that in a region in which there are
irregularity in meiosis. numerous colonies, subject to frequent ex-
Balanced chromosome rearrangements tinction and refounding by stray fertilized
include inversions and translocations of females from the more successful ones.
various sorts. Paracentric inversions, at Since this is also a situation especially fa-
least in Drosophila and some grasshop- vorable to peak-shifts, there should be a
pers, suffer very little reduction in fecun- strong correlation between speciation by
dity. They are important in evolution by the above process and favorable character
permitting segregation of chromosomes, changes.
or at least large blocks, as wholes. They Studies of the differences between closely
permit adaptation to different ecological related species have made it clear that re-
niches and thus increase the adaptability productive isolation usually involves sev-
of the species, but are not involved in spe- eral processes (review by Dobzhansky,
ciation. 1970). It is clear that incipient reproduc-
Pericentric inversions and transloca- tive isolation, whatever its cause, tends to
tions typically lead to significant reduction be made complete, in one way or other,
of the fecundity of heterokaryons, 50% in by natural selection against individuals in-
the case of a typical reciprocal transloca- volved in hybridization.
tion, and thus to incipient speciation. They The general conclusion of this section
may be important in speciation for this is that speciation may take place in a great
reason even though the character changes many different ways that have nothing in
(from position effects) are slight or absent. common except the promotion of repro-
According to M. J, D. White (1978): ductive isolation and hence branching of
"Over 90 percent (and perhaps over 98 the evolutionary process. The effects of
percent) of all speciation events are accom- speciating events on other characters range
panied by karyotypic changes and-in the from zero where the event is merely the
 MACROEVOLUTION-SmFTING BALANCE THEORY 439

extinction of intermediates in a chain of adaptive radiation into other niches. Such

subspecies, to abrupt and great in the case ecological opportunities are, it has seemed
of amphidiploidy, but in general the ef- to me, much the most important cause of
fects on characters are at the microevo- the origin of higher categories (Wright,
lutionary level. 1941a, 1941b, 1945, 1949). The occur-
rence of any particular kind of mutation
The Higher Categories (usually recurrent) is of little significance
The only aspect of the evolution of the in the absence of such an opportunity.
higher categories of which the actual ge- Such an opportunity arises on entry of
netics is fairly clear, is the course of sub- the species into a region in which niches,
stitution in the amino acid sequences of related to its own, are unoccupied. The
several proteins over long periods of geo- remarkable adaptive radiation of families
logic time (Zuckerkandl and Pauling, of marsupials in Australia and of families
1965). The rates of substitution separating of several primitive orders of mammals in
hemoglobin {3 of man from the globin of South America, both in the early Ceno-
the lamprey (1.3 per site per billion years), zoic, are examples. At a lower level is the
hemoglobins a or {3 of several mammals adaptive radiation of the Geospizidae from
from a of a bony fish (carp), and either a a species of ground finch that reached the
or {3 of several mammals with each other, Galapagos Islands (that attracted Dar-
are surprisingly uniform (grand average win's attention) and that of the Drepani-
1.1 per site per billion years) (Kimura, didae from a species of bird (honeycreeper)
1969). that reached the Hawaiian Islands.
This is largely true of substitutions in A similar opportunity is presented to the
the amino acid sequences of cytochrome-c survivors of a catastrophe that has caused
throughout eukaryotic evolution since the extensive extinction of species. The ex-
separation of the higher plants, insects, traordinary adaptive radiation of the or-
lower chordates, reptiles, birds, and ders and families of mammals in the Pa-
mammals from the fungi some 1.3 billion leocene and Eocene following the
years ago and the separation of the higher worldwide extinction of the dinosaurs and
groups from each other (Fitch and Mar- many other forms at the end of the Me-
kowitz, 1970). sozoic is an example. The mammals had
The rates in the nonessential portions existed as a small group for some hundred
of any given protein have been so uniform million years without much differentiation
that Kimura has held that natural selec- before this opportunity occurred.
tion must be ruled out and has proposed Of special importance, however, are the
that the substitutions are the cumulative cases in which evolution along a restricted
consequences of accidents of sampling. line happens to lead to an adaptation that
While most others advocate mass selection turns out to open up an extensive new way
(e.g., Ayala, 1975) and I have proposed of life. The evolution of a motile, tadpole-
the shifting balance process (Wright, 1978), like larva of a certain type of sessile echi-
the near uniformity of rate over a period noderm seems to have opened the way, by
encompassing the origins of kingdoms and means of neoteny, to the active life of the
phyla implies continuity of the evolution- chordates other than tunicates. Later
ary process. modification of the first gill arch made
The evolutionary potentialities of most possible an effective jaw, leading to the
species are probably restricted to very adaptive radiation of the fishes. Modifi-
slow, gradual progress in adaptation to cation of the ventral fins to permit loco-
the single ecological niche which they oc- motion on land and of the swim bladder
cupy, by the occupancy of all related niches for respiration in the air, enabled crossop-
by other species. From time to time, how- terygian fishes to move from one drying
ever, an opportunity is presented for pool to another, a peripheral niche for a
440 SEWALL WRIGHT

fish, but this opened up life on land and ing as well as the others before it was
the adaptive radiation of the amphibia. fixed.
Similarly certain specializing adaptations It would hardly be possible for a typical
later led to the adaptive radiations of the clam to be derived from a typical snail by
reptiles and birds. mutation or a succession of mutations but
A species presented with an opportunity it is not unreasonable to suppose that a
to invade an unoccupied niche would be small protomollusk with a single simple
able to use mutations with more drastic shell, produced a mutant in which this
effects than the quantitative variants used was divided laterally into two and that the
before and to do so sometimes in spite of pelecypods evolved from this mutant type.
rather unfavorable side effects because of All of the classes of mollusks could rea-
the absence of competition. These may in sonably have arisen in somewhat this way.
a sense be considered to be the species All of the phyla of multicellular inverte-
mutations postulated by Goldschmidt but brates and most of their classes probably
it is the ecological opportunity, not mere arose from small rather simple forms, most
occurrence of this sort of mutation, that of them not long, in geologic terms, after
leads the way. the origin of the eukaryotic cell.
It should be added that much more The pattern of evolution thereafter was
drastic mutations may be utilized and fix- probably one of very gradual orthogenetic
ation may be much more rapid if the mu- progress along many lines with occasional
tation is recurrent and comes to be asso- appearance of ecological opportunities of
ciated with otherwise neutral modifiers the sorts discussed above, followed by ex-
that remove its inevitable deleterious side plosively rapid divergence of species in
effects. In all three of the book reviews exploiting these. The group could be con-
referred to earlier (Wright 1941b, 1941c, sidered to constitute a new genus unless
1945), it was noted that the origin of a secondary divergencies raised the level of
higher taxon is expected to be an "explo- the group to the family and tertiary di-
sively" rapid process under the shifting vergencies, perhaps, raised it to the level of
balance theory. This has been reiterated a new order, all within no more than a
in many later papers. moment in geologic time. At some level,
The reorganization required for the or- major divergence ceases, and the species
igin of the highest categories may seem so settle down with only minor differentia-
great that only "hopeful monsters" will do. tion of new species to slow orthogenetic
Here, however, we must consider the size progress (Wright, 1949).
and complexity of the organisms. Such
changes would probably have been im- SUMMARY
possible except in an organism of very The implications of the shifting balance
small size and simple anatomy. I have re- theory with respect to the course of evo-
corded more than 100,000 newborn guinea lution agree in the main with the pattern
pigs and have seen many hundreds of indicated by the fossil record, according
monsters of diverse sorts (Wright, 1960) to Simpson in 1944 (cf. Wright, 1945), re-
but none were remotely "hopeful," all iterated by Gould and Eldredge (1977) in
having died shortly after birth if not ear- their statement quoted in our introduc-
lier. Yet among nine specimens of a small tion: "Punctuated change dominates the
trematode (Microphallus opacus) about 1.5 history of life: evolution is concentrated in
mm long, of which I made serial sections very rapid events of speciation (geologi-
in my first research project (Wright, 1912), cally instantaneous even if tolerably con-
one was highly abnormal in form and had tinuous in ecological time." I would, how-
two large ovaries instead of only one. It ever, substitute the phrase "of character
would probably have been considered a change" for "of speciation." Character
monster if it had been a large complicated change and speciation (in the sense of re-
organism, but it was apparently flourish- productive isolation) are wholly different
 MACROEVOLUTION-SIflFTING BALANCE THEORY 441

phenomena genetically, even though If on the other hand, the species occu-
closely correlated in occurrence. pies a wide range but in part at least only
There is agreement only with the first sparsely and with restricted dispersion,
but not the last part of the sentence also the operation of the shifting balance pro-
quoted in our introduction, from Gould cess leads to gradually improving adap-
(1980): "Evolution is a hierarchic process tation of the species as a whole, by means
with complementary but different modes of successive minor peak-shifts and selec-
of change of its three leading varieties: tive diffusion from them.
within species, speciation and patterns of Still with only a single niche but under
macroevolution." The shifting balance continuously changing conditions, there is
process is a two-level one (selection among continual readaptation largely of the
individuals and among differentiated local treadmill sort, change without progress.
populations), but no difference is assumed A very gradual improvement is, however,
in the rates of minor and major mutation to be expected as the species is shuffled
during the phases of near-stasis and rapid into the higher general regions of the se-
change. The interpretation of these phases lective topography. This occurs even in
under the shifting balance theory is in populations that are effectively panmictic.
terms of differences in ecological oppor- In such a population, the rate of change
tunity. Speciation tends to accompany is approximately according to Fisher's
rapid change both because each of these fundamental theorem except as qualified
processes tends to bring about the other by frequency dependent selection or link-
and because speciation from chromosome age disequilibrium. If, however, popula-
rearrangement and peak-shifts is fa- tion structure permits significant opera-
vored by the same population structure tion of the shifting balance process,
(numerous small colonies subject to fre- readaptation is facilitated by minor peak-
quent extinction and refounding by stray shifts, not allowed for in Fisher's theory.
individuals from the more flourishing col- In cases in which new ecological niches
onies). become available in any of the ways re-
According to the shifting balance theory ferred to, their occupation may require
the determining factor for rapid change, allelic substitutions with major effects.
and the origin of a new higher taxon that Such substitution may occur in spite of
usually accompanies such change, is the imperfect adaptation and the inevitable
presence of one or more vacant ecological deleterious effects of any major change,
niches, whether from entrance of the because of the absence of competition, but
species into relatively unexploited territo- is greatly facilitated if population struc-
ry or from its survival after a catastrophe ture is favorable to peak-shifts, involving
has eliminated other species occupying re- the gene in question and one or more near-
lated niches, or from gradual attainment ly neutral modifiers that tend to eliminate
of an adaptation that opens up a new way the more deleterious of the side effects.
of life. Such major peak-shifts are most likely if
We consider first the course of evolution the major mutation is recurrent and thus
of a species restricted to a single niche becomes available, sooner or later, wher-
(because of occupancy of all related niches ever a favorable modifier or a pair of such
by other species) and living for a long time modifiers, reaches sufficiently high fre-
under relatively unchanging conditions. If quencies for crossing of a saddle that pulls
its population density is great or there is the major mutation to occupancy of the
a wide dispersion of offspring, a state of higher peak.
near-stasis should soon be reached as it Such occupancy tends to give incipient
comes to occupy the most available selec- reproductive isolation, followed by full
tive peak. It cannot move down from this speciation under selection against hybrid-
against natural selection to reach a saddle ization. On the other hand, speciation
leading to a higher selective peak. may come first, because of geographical
442 SEWALL WRIGHT

isolation of a portion of the species or local DAWKINS, R. 1976. The Selfish Gene. New York
establishment of a chromosome rearrange- Univ. Press, N.Y.
DOBZHANSKY, TH. 1970. Genetics and the Evolu-
ment or other cause, and be followed by tionary Process. Columbia Univ. Press, N. Y.
occupancy of the new niche by means of EAST, E. M. 1910. A Mendelian interpretation of
a major peak-shift. In either case, the de- variation that is apparently continuous. Amer.
cisive peak-shift occurs within a species Natur. 44:65-82.
- - - . 1916. Studies on size inheritance in Nico-
and is likely to be accompanied by minor tiana. Genetics 1:164-176.
peak-shifts that improve adaptation. FISHER, R. A. 1928. The possible modification of
The occupancy of new niches, accom- the responses of the wildtype to recurrent muta-
panied by speciation, constitutes the ori- tions. Amer. Natur. 63:115-126.
gin of a new higher taxon. This may mere- - - - . 1929. The evolution of dominance: a reply
to Professor Sewall Wright. Amer. Natur, 63:553-
ly be a new genus but if there is extensive 556.
adaptive radiation into new niches, the - - - . 1930. The Genetical Theory of Natural Se-
array may constitute a new family, new lection 2nd ed. Clarendon Press, Oxford. Dover,
order, or very rarely a new class or phy- Pub!. 1958.
FISHER, R. A., AND E. B. FORD. 1947. The spread
lum. of a gene in natural conditions in a colony of the
It may seem that mutations with im- moth Panaxia dominula. Heredity 1:143-174.
possibly drastic effects would be required FITCH, W. M., AND E. MARKOWITZ. 1970. An im-
for the origins of the higher of the taxa. proved method for determining codon variability
in a gene and its applicability to the rate of fix-
Such origins, however, probably all oc- ation of mutations in evolution. Biochem. Genet.
curred from species, the individuals of 4:579-593.
which were so small and simple in their GOLDSCHMIDT, R. 1940. The Material Basis of
anatomies that mutational changes, that Evolution. Yale Univ. Press, New Haven.
would be complex in a large form, were GOULD, S. J. 1977. The return of hopeful monsters.
Nat. Hist. Magazine, June-July 1977:22-30.
not actually very complex. - - - . 1980. Is a new and general theory of evo-
The final conclusion is that the evolu- lution emerging? Paleontology 6:119-130.
tionary processes indicated by the fossil GOULD, S. J., AND N. ELDREDGE. 1977. Punctuat-
record can be interpreted by the shifting ed equilibria: the tempo and mode of evolution
reconsidered. Paleontology 3:115-151.
balance theory without invoking any HALDANE, J. B. S. 1924. A mathematical theory of
causes unknown to genetics or ecology. natural and artificial selection. Part I. Cambridge
Phil. Soc. 23:19-41.
LITERATURE CITED - - - . 1932. The Causes of Evolution. Harper and
AYALA, F. J. 1975. Scientific hypotheses, natural Brothers, Pub!., N.Y.
selection and the neutrality theory of protein evo- HARDY, G. H. 1908. Mendelian proportions in a
lution, p. 19-42. In F. M. Solzanao (ed.), The mixed population. Science 28:49-50.
Role of Natural Selection in Human Evolution. HEDRICK, P. W. 1981. The establishment of chro-
North Holland Pub!. Co., Amsterdam. mosomal variants. Evolution 35:322-332.
CASTLE, W. E. 1903. The laws of Galton and Men- HUXLEY, J. S. 1942. Evolution, the Modern Syn-
del and some laws governing race improvement thesis. Harper and Brothers, Pub!., N.Y.
by selection. Proc. Amer. Acad. Arts Sci. 39:233- JENKIN, F. 1867. Origin of species. North British
242. Rev. 46:277-318.
- - - . 1916. Further studies of piebald rats and KELLOGG, V. L. 1907. Darwinism Today. Henry
selection with observations on gametic coupling. Holt and Co., N.Y.
Carnegie Inst. Washington. Pub!. no. 241, p. KIMURA, M. 1968. Evolutionary rate at the molec-
163-190. ular leve!. Nature 217:624-626.
- - - . 1919. Piebald rats and selection: a correc- - - - . 1969. The rate of molecular evolution con-
tion. Amer. Natur. 53:370-376. sidered from the standpoint of population ge-
CHARLESWORTH, B. 1979. Evidence against Fish- netics. Proc. Nat. Acad. Sci. USA 63:1181-1188.
er's theory of dominance. Nature 78:848-849. KING, J. L., AND T. H. JUKES. 1969. Non-Darwin-
DARWIN, C. 1859. The Origin of Species by Means ian evolution. Science 164:788-798.
of Natural Selection. 6th ed. John Murray, Lon- MAYNARD SMITH, J. 1975. The Theory of Evolu-
don. D. Appleton and Co., New York. 1910. tion. Penguin, London.
- - - . 1868. The Variation of Animals and Plants MAYR, E. 1963. Animal Speciation and Evolution.
Under Domestication. 2nd ed. John Murray, Lon- Harvard Univ. Press, Cambridge.
don. D. Appleton, London. 1883. NILSSON-EHLE, H. 1909. Kreuzungsuntersuchun-
 MACROEVOLUTION-SmFTING BALANCE THEORY 443

gen an Hafer und Weizen. Lunds Univ. Aarskr. - - - . 1941c. The material basis of evolution (Re-
NF 5:2:1-22. view of book by R. Goldschmidt) Sci. Monthly
NORDENSKIOLD, E. 1928. The History of Biology. 53: 165-170.
Translated by L. B. Eyre. Tudor Publ. Co., - - - . 1945. Tempo and mode in evolution: a crit-
N.Y. ical review. Ecology 28:415-419.
NORTON, H. T. J. 1915. In R. C. Punnett (ed.), - - - . 1948. On the roles of directed and random
Mimicry in Butterflies. Cambridge Univ. Press, changes in gene frequency in the genetics of pop-
Cambridge. ulations. Evolution 2:279-294.
OSBORN, H. F. 1934. Aristogenesis, the creative - - - . 1949. Population structure in evolution.
principle in the origin of species. Amer. Natur. Proc. Amer. Phil. Assoc. 93:471-478.
68: 193-235. - - - . 1960. The genetics of vital characters of the
OSGOOD, W. H. 1909. Revision of the mice of the guinea pig. J. Cell Compo Physiol. 56 Suppl.
American genus Peromyscus, North American 1:123-151.
Fauna Bull. No. 28 Washington, D.C. Bur. BioI. - - - . 1964a. Pleiotropy in the evolution of struc-
Survey, USDA. tural reduction and of dominance. Amer. Natur.
SEARLE, A. G. 1968. Comparative Genetics and 98:65-69.
Coat Colour in Mammals. Academic Press, N.Y. - - - . 1964b. Stochastic processes in evolution, p.
SILVERS, W. K. 1979. The Coat Colors of Mice: a 199-241. In John Garland (00.), Stochastic Models
Model for Mammalian Gene Action and Inter- in Medicine and Biology. Univ. Wisconsin Press,
action. Springer-Verlag, N.Y. Madison.
SIMPSON, G. G. 1944. Tempo and Mode in Evo- - - - . 1968. Evolution and the Genetics of Pop-
lution. Columbia Univ. Press, N.Y. ulations, Vol. 1. Genetic and Biometric Foun-
VRIES, H. DE. 1901-1903. Die Mutations theorie, dations. Univ. Chicago Press, Chicago.
2 Vol. Veit Co., Leipzig. - - - . 1969. Evolution and the Genetics of Pop-
WEINBERG, W. 1908. Ueber den Nachweis derVer- ulations, Vol. 2. The Theory of Gene Frequen-
erbung beim Menschen. Jahresheft Ver. Natur- cies. Univ. Chicago Press, Chicago.
forsch. F. Wiirttemberg 64:368-382. - - - . 1977. Evolution and the Genetics of Pop-
WHITE,M. J. D. 1978. Modes of Speciation. W. H. ulations, Vol. 3. Experimental Results and Evo-
Freeman and Co., San Francisco. lutionary Deductions. Univ. Chicago Press, Chi-
WILLIAMS, G. C. 1966. Adaptation and Natural cago.
Selection. A Critique of Evolutionary Thought. - - - . 1978. Evolution and the Genetics of Pop-
Princeton Univ. Press, Princeton. ulations, Vol. 4. Variability Within and Among
WRIGHT, S. 1912. Notes on the anatomy of the tre- Natural Populations. Univ. Chicago Press, Chi-
matode, MicrophaUus opacus. Trans. Amer. Micr. cago.
Soc. 31:167-175. - - - . 1980. Genic and organismic selection. Evo-
- - - . 1929a. Fisher's theory of dominance. Amer. lution 34:825-843.
Natur. 63:274-279. YULE, G. U. 1902. Mendel's laws and their prob-
- - - . 1929b. Evolution in a Mendelian popula- able relation to interracial heredity. New Phytol.
tion. Anat. Rec. 44:287. 1:192-207, 222-238.
- - - . 1931. Evolution in Mendelian populations. - - - . 1906. On the theory of inheritance of quan-
Genetics 16:97-159. titative compound characters and the basis of
- - - . 1932. The roles of mutation, inbreeding, Mendel's laws: a preliminary note. Proc. ill In-
crossbreeding and selection in evolution. Proc. temat. Congo Genet. p, 140-142.
VI Intemat. Genet. Congo 1:356-366. ZUCKERKANDL, E., AND L. PAULING. 1965. Evo-
- - - . 1940. Breeding structure of populations in lutionary divergence and convergence in pro-
relation to speciation. Amer, Natur. 74:232-248. geins, p. 97-166. In V. Bryson and H. J. Vogel
- - - . 1941a. On the probability of fixation of re- (eds.), Evolving Genes and Proteins. Academic
ciprocal translocations. Amer. Natur. 75:513-522. Press, N.Y.
- - - . 1941b. The "Age and Area" concept ex-
tended. (Review of book by J. C. Willis) Ecology Corresponding Editor: D. J. Futuyma
22:345-347.