01 Understanding Natural Selection

Cambridge Books Online
http://ebooks.cambridge.org/
Evolutionary Game Theory, Natural Selection, and Darwinian Dynamics
Thomas L. Vincent, Joel S. Brown
Book DOI: http://dx.doi.org/10.1017/CBO9780511542633
Online ISBN: 9780511542633
Hardback ISBN: 9780521841702
Paperback ISBN: 9781107406513
Chapter
1 - Understanding natural selection pp. 1-25
Chapter DOI: http://dx.doi.org/10.1017/CBO9780511542633.002
Cambridge University Press

1
Understanding natural selection
The following observations about patterns in nature have captured the imagi-
nation of humans for millennia.
1. Fit of form and function (FF&F): different organisms appear remarkably
well suited and engineered for their particular environments. The high-
crowned molars of zebras and white rhinoceros act as mulching mowers
for grinding grass, and protect against the inevitable wear imposed by the
silica content of grass. Black rhinos, on the other hand, have lower crowned
molars favoring efficient mastication of leaves and foliage. None of these
animals has the sharp and stabbing canines like those of lions. Distinct
species1 of organisms apply themselves to different ecological tasks using
their appropriate sets of tools. For example, zebras and white rhinoceros
feed on grass, black rhinos browse leaves from shrubs, and lions kill and
eat zebras.
2. Diversity of life: we share this planet with a phenomenal array of different
life forms. These forms range from delicate mosses and annual flowering
plants to awesome whales and fearsome sharks. While many of these forms
differ in subtle ways, most can be readily recognized and categorized as
types or species quite distinct from others. This is possible because the
extant denizens of our planet do not exhibit a continuum of morphological
variation from bacteria to redwood tree. Rather, the morphologies and
characteristics of living organisms cluster like conspicuous and discrete
galaxies in morpho-space.
3. Procession of life: despite the variety and discreteness of life, organisms
seem connected by design rules of increasing levels of complexity. Notions
such as the tree of life identify a regular, yet increasing, sophistication of
organisms in terms of size, behavior, and the number and specialization of
1 A formal definition of species is given in Subsection 8.2.2.
Cambridge Books Online © Cambridge University Press, 2009

Downloaded from Cambridge Books Online by IP 130.237.36.154 on Thu Mar 31 09:33:41 BST 2016.
http://dx.doi.org/10.1017/CBO9780511542633.002
2 Understanding natural selection
traits. The early idea of a bauplan recognized the fixity of certain design
rules among definable groups of species. Linnaeus in his binomial nomen-
clature used design rules to place organisms in the tree of life. Modern
systematics and taxonomy, now more than ever, rely on the hierarchical
structuring of traits among collections of species to assign names and po-
sition within life’s tree.
4. Distribution and abundance of organisms: this is the central question
of ecology. Paleolithic peoples probably pondered this as the central ques-
tion of survival. Organisms are not spread randomly in space and time.
Furthermore, some organisms seem ubiquitous and excessive in numbers
(various species of crow, for their size, are particularly abundant around
the globe) while others puzzle us with their rarity (the introduced Eurasian
tree sparrow has a toe-hold in the city of St. Louis while its congener, the
European house sparrow, occupies the rest of North America).
These observations must predate recorded history. Yet a satisfactory and
unified answer to why the above four patterns exist has been available for
only about 150 years with the development of Darwin’s theory of evolution
by natural selection. More recently, game theory (the mathematics used to
study conflicts of interest among two or more players) – is being successfully
applied to modeling natural selection. The classical game theory of economics,
sociology, and engineering has existed as a formal discipline since the 1940s
and 1950s, while game theory as a formalism for natural selection has existed
since the 1970s.
The objective of this book is to show that the synthesis of Darwin’s ideas
within the context of an evolutionary game provides a most useful tool for
understanding the four patterns of nature. Because the use of evolutionary game
theory to model natural selection requires a moderate amount of mathematics,
we provide all of the concepts and mathematical tools needed in the chapters
that follow.
In this chapter, we start by discussing Darwin’s marvelous idea of natural
selection, introduce life as an evolutionary game, and explain why we favor a
game theoretic approach as a complement to the more familiar and orthodox
genetical approaches to natural selection.
1.1 Natural selection

1.1.1 Historical perspective
It is appropriate that well into the Age of Enlightenment the field of evolutionary
ecology resided within the intellectual pursuit of Natural Philosophy. Natural

1.1 Natural selection 3
Philosophy encompassed all aspects of the sciences. Then, as today, philosophy

(literally the love of wisdom) pursues the facts and principles of reality. Ecology
falls into this quest for understanding Nature’s reality, and natural philosophers
recognized a wisdom to nature. All organisms exhibit in their characteristics
excellent engineering in fit of form and function (FF&F) and the engineering
shows a commonality and connectedness of design across all life from simple
to complex (procession of life). It is remarkable that, over the ages, the diverse
natural philosophies have all recognized a design and engineering component
to nature. And, until the mid 1850s, all of these philosophies drew a very
logical connection between human tools and organisms as tools designed by
nature.
The connection between the tools in the human household and organisms
in nature’s house is compelling. Hence, essentially all pre-Darwinian natural
philosophies took the next logical step. Tools exist because humans design
and fashion them with purpose and forethought of intent, a watch is proof of
a watchmaker.2 Commonality of features among watches reflects the watch-
maker’s trademark and level of technology. It then follows that biodiversity is
a reflection of a Creator, of gods, of Mother Earth, or of some other personified
force that shows intent and purpose in the conscious design of its organisms.
For most cultures over most of history this logical construction held sway. Just
as humans make tools so something greater (singular or plural, masculine or
feminine) made life. This philosophical view of life provided a seamless blend
for people’s ecological knowledge and spiritual beliefs. In the nineteenth cen-
tury (Darwin’s Century as Eiseley (1958), aptly calls it) in Western Europe,
and in England in particular, this viewpoint began to lose favor as applicable to
biology.
Lyell’s geology showed how ongoing forces and non-personified natural
processes could explain the forms, types, and layering of rocks (Lyell, 1830).
And within many of these distinctly non-living rocks were the distinct re-
mains of previous life. Erosion, sedimentation, compression, and volcanism
provided for geological changes with time. Could the fates of rocks and life
be tied together? Could similarly non-personified natural forces explain the
origins and changes of life with time? The essentialists (linked to Greek
ideas of life mirroring or manifesting some deeper fixed reality and truth) and
biblical creationists (Genesis as scientific treatise) scrambled to make sense of
2 Apparently William Paley was the first to use the analogy. “. . . suppose that I had found a watch
upon the ground . . . this mechanism being observed the inference we think is inevitable, that the
watch must have had a maker . . . or shall it, all at once turn us around to an opposite conclusion,
namely that no art or skill whatever has been concerned in the business. Can this be maintained
without absurdity?” Evidence of evolution reveals a universe without design, hence Dawkins’s
(1986) useful metaphor of the blind watchmaker.

these new findings and ideas. In its more complex forms, scientific creation-
ism stretched biblical days into millennia, and recognized multiple creations
and destructions of life, of which Noah’s Flood was but one particularly
noteworthy example (Schroeder, 1997). But those seeking a “uniformitarian”
explanation for life also had major conceptual and logical hurdles. Yes, ge-
ology and life seemed to share a common fate, but erosion, sedimentation,
and volcanism do not form the characteristics of organisms. Empirically,
life might change its characteristics with time, but what were life’s natural
processes?
Evolution built around heritable change with time was a potentially attrac-
tive force. Most natural philosophers accepted the presence of this force within
animal and plant breeding, and many social philosophies emphasized the con-
nections between human bloodlines and human hierarchies. But, as a force for
change, it was presumed to be rather limited and in most cases useful only for
protecting good blood from bad. Few saw breeding as providing the force or
opportunity for truly novel evolutionary change. Early attempts at linking evo-
lution to FF&F and procession of life still clung to the notion of foreordained
or consciously driven improvement. Some espoused a kind of creationist–
evolutionist blend: a view that saw God creating life at all levels followed by the
evolution of these forms up a chain of being towards humans, angels, and be-
yond. Lamarck advanced a tenable theory of evolution via “self improvement.”
Just as an individual can be conditioned physically for a task, perhaps a species
can condition their heritable characteristics towards needs and particular tasks,
leading to the inheritance of conditioned or acquired traits. Two aspects of this
theory of evolution are interesting. First, Darwin did not see Lamarck as in-
compatible with natural selection and in fact viewed the inheritance of acquired
traits as one of several likely ways for introducing heritable variation. Second,
Lamarckism could have been correct as a scientific perspective. If pangenesis
(the equal contributions of all units of the body to the heritable blueprint for the
organism’s offspring) had been correct, then acquired (or discarded) traits could
manifest as heritable change, and natural selection could work within this con-
text. And indeed, in prokaryotes, and some plants where there are fewer clear
boundaries between the somatic cell line and the gametic cell line, manifesta-
tions of Lamarckian evolution do occur comfortably within the framework of
natural selection. But, the raindrops that eroded and formed Lyellian geology
still eluded evolutionist thinking.
Darwin found the raindrops in deceptively simple ecological processes –
surplus births and subsequent famine. The Struggle for Existence (loosely
associated with Malthus (1796)) recognizes a reality of ecology. Organisms
are capable of having many more offspring than the environment can possibly

1.1 Natural selection 5
support. Darwin’s genius was in making the link between heritable variation
(however it came about!) with the Struggle for Existence in which less satisfac-
tory individuals die. Just as raindrops sculpt landscapes by eroding softer and
harder stones at different rates, the ecological raindrops of births and deaths
striking the softer and harder rocks of heritable characteristics sculpt life. It is
not hard to see how many a natural philosopher would find repugnant the deep
social irony of natural selection as beautifully described in “Darwin’s Danger-
ous Idea,” (Dennett, 1995). This repugnance resonates today in the writings of
intellectuals such as Gould (1998). The “noble” excellence exhibited by FF&F
and procession of life is engineered by the scourges put upon it as manifested
by “poverty” and “famine.”
1.1.2 As Darwin saw it

Evolution is the physical, genetic, or behavioral change in populations of
biological organisms over time. Evolution’s more interesting and significant
manifestations result from natural selection, a process that engineers bio-
logical systems. Natural selection works within genetic, developmental, and
environmental constraints to shape biological organisms in ways that make
them appear adapted to their environments. Understanding an evolutionary de-
sign has its roots in Darwin’s postulates (Darwin, 1859). As Sober (1984,
p. 21) notes, Darwin’s postulates are really two drawn out, discursive propos-
itions. Darwin saw heritable variation leading to evolution, and evolution lea-
ding to new species and to new distributions of characteristics within species.
Drawing from Lewontin (1974), we will separate Darwin’s argument into three
postulates:
1. Like tends to beget like and there is heritable variation in traits associated
with each type of organism.
2. Among organisms there is a struggle for existence.
3. Heritable traits influence the struggle for existence.
The first postulate was generally well known at the time and had been used by
plant and animal breeders for centuries to improve native strains. The second
postulate was influenced by Malthus’s Essay on Population (1976) with
the thesis that resources can only increase arithmetically while human pop-
ulations grow geometrically. Darwin extended this idea into the general
phenomenon of competition among individuals of the same or different
species for limiting resources. Darwin’s last postulate provided the key
for understanding the consequences of evolution. For a particular environ-
ment, this postulate results in an increase in phenotypically well endowed

individuals who are better able to survive and reproduce than less well endowed
individuals.
Darwin used logical verbal arguments to model evolution. His views on
inheritance were both orthodox for the day and flawed. Today, we think of evo-
lution in terms of genetics, which involves the study of inheritance of genes
from one generation to the next. Genetics seems to provide the ultimate tool for
studying evolution, yet it is a curious fact that Darwin presented his theory in
the absence of any understanding of genes as presented by Mendel (1866). It
was not until the 1930s that Fisher (1930), Wright (1931), Haldane (1932),
Dobzhansky (1937), and others combined evolution and genetics into what is
known as the Modern Synthesis (Mayer and Provine, 1980). Genetics has pro-
vided a framework for understanding evolution, yet it need not be the essential
core for modeling or understanding evolution by natural selection. Darwin’s
postulates do not require any specific mechanism of inheritance. This obser-
vation is in accordance with the development presented in this book. Since
Darwin’s three postulates constitute a fundamental principle that can be used
to explain and predict evolution, we use these principles in developing a non-
genetical mathematical framework for natural selection. The framework is not
non-genetical in the sense of not having some mechanism for inheritance, and
an understanding of the recipe of inheritance, as in the case of modern genetics,
is paramount to Darwin’s first postulate (as well as to bioengineering, medical
genetics, animal and plant breeding programs, taxonomy, DNA fingerprinting,
etc.). The framework is non-genetical in the sense that an actual genetic system
for allowing natural selection is an auxiliary hypothesis. In the same manner
natural selection is merely an auxiliary hypothesis (among several evolutionary
forces) for changes to a genetic system. We propose that evolution by natu-
ral selection is a dynamic game. Our objective is to develop an evolutionary
game theory that can be used as a fundamental modeling tool for understanding
natural selection.
1.1.3 The Modern Synthesis

The Modern Synthesis that began in the 1900s and was completed by the 1930s
is often viewed as a critical step in formalizing natural selection (Sober, 1984).
The lack of a mechanism for inheritance hampered development of rigorous
mathematical models of natural selection, which in turn hampered applic-
ation and advancement. The “rediscovery” of Mendel’s Laws in the 1900s
(Pearson, 1904; Hardy, 1908) energized work on breeding and inheritance, and
drew into question the compatibility of Mendel’s particulate inheritance with

1.2 Genetical approaches to natural selection 7
natural selection. Fisher (1930), Wright (1931), Haldane (1932), and others ush-
ered in a golden age of population genetics by placing the study of evolution on
a firm mathematical foundation. In creating this foundation, they showed the
compatibility of Mendelian genes, loci, and alleles with natural selection, the
evolution of quantitative traits, and systematics. In addition, the recipe of inheri-
tance provided insights into other forces of evolution (mutation and genetic drift)
and into interactions that might occur genetically within and between organisms.
Genetic interactions within an organism could be epistatic (many genes at dif-
ferent loci may contribute non-additively to a particular trait) and pleiotropic (a
single gene may contribute to the phenotype of several traits). Among individu-
als, natural selection could be density dependent and/or frequency dependent
depending on whether the population’s size and/or gene frequencies influence
the success of individuals with particular phenotypes, respectively.
The Modern Synthesis led to the primacy of genes over heritable phenotypes
as the objects of evolution. This primacy seems self evident. In the Modern
Synthesis, evolution is defined as a change in gene frequency. However, nat-
ural selection in terms of FF&F must involve the ecological consequences of
heritable phenotypes. Can a strictly genetical approach be sufficient for model-
ing natural selection? Models of gene-frequency dynamics determine what has
been selected but cannot necessarily determine what survival or fecundity apti-
tudes of the organism have been selected for. The FF&F requires understanding
both what has been selected and why. The “why” requires a focus on heritable
phenotypes, particularly when natural selection is frequency dependent. So,
while the Modern Synthesis provided a huge advance in our understanding of
evolution, taxonomy, and gene dynamics, it may have unwittingly hampered a
fuller appreciation of natural selection by subordinating heritable phenotypes
to their genetic recipes.
1.2 Genetical approaches to natural selection

Population genetics (modeling changes in the frequency of particular alleles
within a population) and quantitative genetics (modeling the change with time
of quantitative traits under the assumption that many alleles and loci contribute
more or less additively to the trait value within an interbreeding population)
are the concepts currently used for thinking about and modeling evolution
where evolution is defined as a change in gene frequency. This outlook guided
research to examine how genetic variability and genetic constraints direct and
restrict evolutionary change (Crow and Kimura, 1970).

Viewing evolution as change in gene frequency can produce reasonable re-

sults in terms of producing an FF&F. For example, consider the case where
the fitness conferred by a gene on an individual is density independent
(independent of the population size) and frequency independent (indepen-
dent of gene frequencies). In this case, the gene dynamics favor the genes that
confer the highest per capita rate of growth on the population. In the situation
where the fitness conferred by a gene is density dependent and frequency in-
dependent, then gene dynamics favors genes that maximize the population’s
size. In both of these cases the gene dynamics favors survival of the fittest if
fitness is defined either as population growth rate or population size. However,
as soon as evolution is frequency dependent, that is the fitness conferred by
a gene on an individual is influenced by the frequencies of other genes in the
population, then the linkage between the consequence of natural selection op-
erating on genes and some corresponding measure of fitness at the population
level disappears. The endpoint of the gene dynamics no longer optimizes any
obvious measure of ecological success. This will be the most common situation
as plausible genetic interactions such as epistasis, pleiotropy, and heterozygote
superiority all introduce frequency dependence. The decoupling of change in
gene frequency from some measure of ecological success for the individual
organism or the population has unintended and unfortunate consequences for
the question of FF&F. When evolution by natural selection becomes simply the
endpoint of genetic dynamics, evolution by natural selection becomes poten-
tially tautological. The fittest genes are those that survive and so survival of
the fittest becomes a truism. Or it encourages a view of a life in which genes
are the engineers of blindly programmed robots that serve only to reproduce
more genes (paraphrased from The Selfish Gene (Dawkins, 1976)). The wings
of a bird are no longer for flying; rather they are a part of the machinery for
proliferating genes. The FF&F concept is lost in favor of the dynamical system
of gene frequencies.
In this book, the focus will be on the wing rather than the genes coding
for the wing. Characters such as wings will be modeled as evolutionary strat-
egies (heritable phenotypes). Even under frequency-dependent selection, the
resulting game theory analysis will reveal both what has been selected and
why. The FF&F requires us to study strategies as the outcome of an evo-
lutionary process (accessible using gene-frequency dynamic models), and to
study strategies by their function (tricky when using strictly genetical mod-
els of frequency-dependent selection). A game theoretic approach is needed
because frequency-dependent selection is ubiquitous in natural selection and
plays the key role in the diversity of life and the distribution and abundance of
organisms.

1.2 Genetical approaches to natural selection 9
A consequence of the strictly genetical approach used in current textbooks

on evolution3 is a narrow perspective on genetic variability and a decoupling of
the concepts of microevolution (small evolutionary changes) from concepts of
macroevolution (large evolutionary changes and the stuff of the procession of
life). In evolution courses, such traits as tongue rolling and blood type serve to
emphasize the idea of genetic variability. Once the genetic variability has been
identified, the loci and alleles specified, and the consequences of genes for sur-
vival and fecundity defined, then population genetics brings mathematical rigor
to subsequent changes in gene frequencies brought about by natural selection.
Unwittingly though, the focus on extant genetic variability greatly reduces our
appreciation of the complete set of heritable variation on which natural selec-
tion operates. Subsequent analyses give the impression that natural selection
is a finishing school for microevolution but is inapplicable to macroevolution.
Natural selection becomes subordinated to the known and accepted machinery
of population and quantitative genetics which then gets subordinated to explain-
ing readily observable evolutionary changes within populations. By not being
able to apply the genetical approach to the big interesting evolutionary changes
that separate species, families, orders, and classes from each other, evolutionists
have proposed macroevolutionary forces such as genetic revolutions, species
selection, and phylogenetic constraints and inertia that have little grounding
in natural selection (Eldridge and Gould, 1972; Stanely, 1979; Vermeij, 1994).
Current evolutionary teaching reflects this split in intellectual thinking. The
rigors of population and quantitative genetics are used to show how natural
selection can shape characteristics of populations, and then this machinery is
discarded and replaced when the course moves on to the really interesting ques-
tions of speciation, biogeographic patterns, and the evolution of characters that
define and separate the higher taxa of life. Because macroevolution does not fit
comfortably within population genetics, natural selection becomes separated
from the question of the diversity of life and the procession of life by virtue of
its association with genetical models.
Genetical views of natural selection often ignore the most appealing appli-
cations of natural selection to FF&F, diversity of life, and procession of life.
This happens because a genetical basis for natural selection cannot comfortably
account for the seemingly limitless, though constrained, set of heritable vari-
ability available to natural selection, and it subordinates the organism’s ecology
to the genetic mechanism. But in Darwin’s original formulation it is the eco-
logical interactions operating on the set of evolutionarily feasible phenotypes
3 Frequency-dependent selection often gets short shrift in these textbooks. Usually the most
interesting examples of natural selection cited involve frequency dependence (at least implicitly)
even as the formalisms for conceptualizing frequency dependence receive minimal attention.

that sculpt and refine species towards an FF&F. This aspect of Darwin’s pers-
pective on natural selection represents an adaptationist research program
which studies the advantages that particular characters might confer on the
individual. Fields such as physiological ecology, functional morphology, and
behavioral ecology (particularly in the guise of foraging theory and socio-
biology) produce more or less plausible hypotheses for the adaptation of an
organism’s heritable traits.
The adaptationist approach to natural selection is appealing in that it seems
to contain the spirit of Darwin’s original idea. However, it is built on a poor
foundation. As scathingly noted in “The Spandrels of San Marco” (Gould and
Lewontin, 1979), the intuitively appealing explanations for the value of traits
to an organism rested on non-rigorous and often indefensible notions of what
is valuable to an organism and what is heritably feasible. The adaptationist
paradigm in the 1970s lacked formal fitness functions, formal statements of what
was feasibly heritable, and formal evolutionary dynamics. Here’s the dilemma.
Genetical approaches have been successful at modeling what is selected but
lack insights into why a character has been selected. Adaptationist approaches
have been successful at proposing why a character has been selected for, but
often lack a modeling framework.
Here we take another look at the adaptationist approach as embodying the
spirit of Darwin’s theory of natural selection. While we applaud the formalism
and rigor of population genetic and quantitative genetic approaches to evolution,
we regard life as a game, and that a game theoretic approach provides the right
tools and a sufficient level of rigor for an adaptationist approach to evolution
by natural selection. In this book, we present life as a game and develop the
formalism necessary to model evolution by natural selection as a game. To make
the transition from a strictly genetical perspective to a game theoretic one, we
view evolution as a change in heritable phenotypes rather than as a change in
gene frequency. From this viewpoint, we recover the sense of natural selection
as an optimization process leading to adaptations, and support the engineer’s
perspective that organisms are designed for a function.
1.3 Natural selection as an evolutionary game

The long loop of Henle within a kangaroo rat’s kidney allows it to produce
exceedingly concentrated urine. Because of this and other physiological adap-
tations (Schmidt-Nielson, 1979), the kangaroo rat can inhabit deserts, eat little
more than seeds, and never drink a drop of water in its lifetime. Mussels in-
habiting inter-tidal habitats have strong abyssal threads that lash them to the

1.3 Natural selection as an evolutionary game 11
rocks and prevent them from being swept away by crashing waves. At low
tide they clam up tight to prevent desiccation and at high tide they open up
to filter food particles from the swirling water. These physiological, morpho-
logical, and behavioral traits seem sensible in the organisms’ ecology. They
meet challenges and exploit opportunities. These traits must be heritable, and
natural selection has produced this FF&F. The kangaroo rat’s kidney and the
mussel’s threads represent evolutionary design features that allow these animals
to survive in a seemingly optimal way under the circumstances. Explanations
for these adaptations fall squarely within Darwin’s theory.
For some species of frogs, choruses of males sing at ponds and waterways
in order to attract females. In Costa Rica, for instance, the calls also attract the
attention of frog-eating bats (Tuttle and Ryan, 1981). The males, as in many
species of the animal world, are literally dying for love. A chorusing male
calls for a mate, but may call in its predator. Yet it still choruses. Why? And,
more fundamentally, is this adaptation consistent with FF&F? Such chorusing
is not the only way for females to find mates. There are other frog species that
achieve match-making more quietly. There are costs and benefits to chorusing,
such calling seems to make it easier for females to find mates, allows males
to advertise their presence, reduces male survivorship, and feeds bats. A col-
lective reduction in calling volume by the males would probably hamper bats’
effectiveness as predators with negligible effect on females’ access to mates.
This apparent paradox in behavior is resolved when mating is viewed as a game
played with many other frogs. The male frog’s best chorusing strategy depends
upon the strategy choice of females and the calling strategy of other males. In
the frog’s mating game, chorusing functions primarily to attract mates away
from other males. An economist would quickly see this mating system as an
advertising game, where advertising expenditures serve primarily to divert
customers from one’s competitors rather than to increase the overall pool of
customers.
Given the strategies used by the females and other males, the male’s chorus
strategy is optimal in the sense of maximizing his reproductive success that
depends on the product of survivorship and mating success. Suppose that mating
success increases and survivorship declines with the volume of an individual’s
chorusing; and that the reverse happens in response to the chorusing volume of
other males. If most males chorus quietly, it behooves a male to call louder –
it gains more from mating success than it loses in survivorship. If most males
are exceptionally loud, it may behoove the male to call more quietly – it gains
more from survivorship than it loses from fewer mating opportunities. Between
these advertising extremes, a best strategy exists where the male calls at the same
volume level as all of the other males. The whole business of calling loudly

may seem maladaptive but, given the circumstances, the calling behavior of the
individual male frog maximizes his reproductive success.
There are similarities in other situations that, at first glance, seem radically
different from the perspective of the organisms’ natural histories. For instance,
what does the evolution of height in trees have to do with chorusing in male
frogs? They are fundamentally the same evolutionary game! Woody plants in-
vest resources (photosynthate) into non-productive wood that presumably could
have been used for reproduction or the production of seemingly more useful
structures such as leaves and roots. The trunks of trees precipitate all sorts
of additional challenges for the plant: susceptibility to tipping over in high
winds, greater surface area for the invasion of pathogens and herbivores, hy-
drostatic problems associated with nutrient exchange between roots and leaves,
and weight and balance issues associated with supporting tall, narrow struc-
tures. Trees behave as trees for the sake of sunlight. Since the pool of available
sunlight is fixed, a tree grows to escape the shade of other trees. At the op-
timum height, being taller than the others around you incurs greater costs in
height than benefits in light. Being shorter than others incurs greater losses in
light than benefits in less height. Tree trunks function primarily to gather light
away from other plants. In the presence of canopy trees, all striving to be shade
free, there exist opportunities for a diversity of light-gathering strategies. Light
gaps provide opportunities for fast-growing, light-loving shrubs and herbs. The
seasonality in temperate zones provides spring windows of light for early flower-
ing herbs that grow and flower before the canopy trees have had time to leaf
out. The flecks of light that pass through the sieve of canopy leaves provide
opportunities for shade-tolerant plant species. The game of light competition
may not only select for tree trunks (FF&F), but also select for a co-existing
diversity of strategies (diversity of life).
Two important properties are essential elements of any game. First, each
player’s success depends not only on its own strategy but on the strategies
of the other players. Second, the players’ objectives are generally at cross
purposes; that is, the combination of strategies among players that is preferable
to one player need not be the arrangement most preferable to any of the other
players.
Nature abounds with games. The frogs and trees engage in different forms
of the tragedy of the commons (Hardin, 1968) in which gains to individuals
at the expense of the group encourage organisms (and humans) invariably to
overuse their common pool of resources. Pursuit-evasion games occur within
most, if not all, predator–prey systems. Arms races occur in the competition
for mates, food, space, and safety. Games of chicken occur in many example of

interference competition. The prisoner’s dilemma characterizes many games

involving the evolution of cooperative behaviors.
All of life is a game. Organisms are not masters of their own fate. In evolving
favorable characteristics, no organism is free from the evolutionary meddling
of others. Indubitably, the advantage or functioning of many heritable traits can
only be understood in the context of the traits of others, that include the same
and/or different species. From the perspective of a game, the strategies used by
trees and frogs can be not only understood, but actually predicted.
Darwin recognized life as a game. In describing sexual selection, he saw
that many behavioral and morphological traits of sexually reproducing animals
only made sense in terms of mate competition. The peacock’s feathers, male
leks, and elaborate courtship rituals, rather than functioning directly for sur-
vivorship or food, serve the suitor in terms of quality and quantity of mates.
Altruism, traits that seem to provide “public” goods to others at a “private”
cost to the individual, was troubling for Darwin and early proponents of natural
selection. Enlightened self-interest provided clues. Costly but nice behaviors
may yield the individual indirect or delayed rewards from within its social con-
text. While recognizing the challenge of some of nature’s games for natural
selection, Darwin and his contemporaries did not possess the formal tools of
game theory to logically and deductively assess the traits whose adaptive signi-
ficance lies in the traits of others.
Wright (1931) (see also Wright, 1932) sought the optimality of traits sup-
ported by natural selection. His construct of the fitness landscape, where a
measure of fitness is plotted against gene frequency, provided a visual repre-
sentation of adaptations as peaks of these landscapes. Fisher’s Fundamental
Theorem of Natural Selection (Fisher, 1930) provided a dynamic for how nat-
ural selection would drive gene frequencies to these peaks. He showed how the
rate of evolution in response to natural selection proceeds up gradients of higher
fitness and at a rate proportional to the product of heritable variability (addi-
tive genetic variability) and the magnitude of the fitness gradient. This aspect
of natural selection as an optimization process that climbs the slopes of fitness
landscapes worked well for density-independent and frequency-independent
selection. In these cases, the landscape is rigid in response to gene frequencies
within the population. Gene frequencies change, but the shape of the land-
scape does not. Under both frequency dependence and density dependence, the
landscape is no longer rigid in response to changes in gene frequencies, and
optimization of fitness is no longer applicable. Because Wright’s analysis came
before the development of game theory, formal analysis of a flexible landscape
was not possible.

Fisher4 in addressing the evolution of sex ratios recognized gender frequen-

cies as a game and flirted with tools of game theory. He saw how, in diploid,
sexually reproducing organisms, the ultimate fitness payoffs to each gender
must be equal. Furthermore, a mother producing offspring achieves fitness in
her grandchildren through both her sons and her daughters. Sons enter a kind
of lottery for half of the genetic payoffs to the next generation, while daughters
join a lottery for the other half. This occurs whether the pool of individuals in
the next generation is fixed, or is determined solely by the number of females,
or some combination of males and females. If the male side of the lottery is
oversubscribed it is best to produce daughters, and vice versa in producing sons.
Hence, it is best to devote equal reproductive resources towards the production
of male and female offspring. In formalizing this centerpiece of sex-ratio theory,
Fisher (1930) anticipated Maynard Smith’s (1976) ESS concept and Bishop and
Cannings’s (1976) general result that, in a matrix game with a mixed-strategy
solution, each pure strategy of the ESS will yield equal payoffs to the player.
Hamilton (1963) in his contribution to the evolution of cooperation by kin
selection touched upon optimality as a game. He formulated the idea that the
advantage of cooperative behavior may lie in its inclusive fitness effect on rela-
tives. Haldane (1932) anticipated the idea of inclusive fitness when he observed
that a person’s self-sacrificing strategy would actually benefit if the person were
willing to sacrifice his life to save three of his brothers who are presumed to have
a fifty–fifty chance of sharing this strategy. The likelihood of like interacting
with like (interactions among relatives providing such a context) can render al-
truistic behaviors adaptive. Altruism can function to maximize an individual’s
genetic prospects given circumstances of non-random interactions.
Levins (1968), with his concept of fitness sets and evolution in hetero-
geneous environments, provided a theory of evolution by natural selection
based on strategies (i.e., heritable phenotypes). He also advanced the approach
of using heritable phenotypes to see how natural selection may promote di-
versity. He first proposed a strategy set for a quantitative trait. The strategy
set represents all of the evolutionarily feasible values for the strategy. He then
assumed that the environment by virtue of heterogeneity offers different oppor-
tunities to the organism. Next, he reasoned that natural selection will always
favor strategy values that simultaneously improve an organism’s prospects in
both habitats of the heterogeneous environment. Those remaining strategies
that trade-off performance in the two habitats become the active edge of the
4 Among Fisher’s contributions to genetics and statistics is his measure of indeterminacy, now
called Fisher information. This concept now plays an important role in theoretical physics
(Frieden, 1998).

fitness set (Levins, 1968). The active edge has similar properties to the Pareto-
optimal solution from game theory. A Pareto-optimal solution has the property
that it is not possible to change strategies among individuals so as to maintain
or increase everyone’s payoff (Vincent and Grantham, 1981). Levins then fixes
the environmental context with respect to the scale (fine- versus coarse-grained
for small vs. large patches of habitat, respectively) and frequency of the two
habitats. One can then solve for an evolutionarily stable strategy by pitting the
active edge of the fitness set against the environmental circumstances. This
produces an optimum that may favor either a single generalist strategy or two
specialist strategies. These strategies produce the highest per capita growth rate
(fitness) given the circumstances.
Ever since Darwin, natural selection has been viewed as an optimizing pro-
cess. One that promotes heritable phenotypes (strategies) that are optimal in
the sense of being the best (maximizing fitness) given the circumstances. How-
ever, when the circumstances include the strategies used by others, evolution
can no longer be viewed in terms of a simple optimization process, rather
it is a game. As a game, natural selection combines evolutionary principles
of inheritance with ecological principles of population interactions to pro-
duce what Hutchinson (1965) called the ecological theater and evolutionary
play.
1.3.1 Game theory and evolution

The genesis of a formal theory of games can be traced to the publication of
Theory of Games and Economic Behavior by von Neumann and Morgenstern
(1944). Game theory had its beginnings with the two-player matrix game, which
introduced the concepts of conflict, strategy, and payoff. There continues to be
a large and growing literature on the theory and application of matrix games
in economics, particularly from an evolutionary perspective (Samuelson, 1997;
Weibull, 1997).
Games are generally divided into three major classes: matrix games, contin-
uous static games, and differential games. Matrix games have a finite number
of strategy choices. After each player makes a choice, the payoff to each player
is determined by an element of a matrix. One player’s strategy corresponds to
the selection of a row and the other player’s strategy corresponds to the selec-
tion of a column. All possible payoffs are given by the elements of the matrix.
The particular payoff a player receives is determined by this combination of
strategies. In continuous static games, the strategies and payoffs are related in
a continuous rather than discrete manner (Vincent and Grantham, 1981). The
game is static in the sense that an individual’s strategy is constant. Differential

games (Issacs, 1965) are characterized by continuously time-varying strate-

gies and payoffs with a dynamical system governed by ordinary differential
equations.
In a conventional game, a rational player’s objective is to choose a strategy
that maximizes his or her payoff. When an individual’s payoff is a function
not only of his or her own strategy, but the strategies of other players, we have
the conditions that define and separate a game problem from an optimization
problem. A game consists of players, strategies and strategy sets, payoffs, and
rules for determining how the strategies employed by the players result in
their respective payoffs. Unlike optimization theory, which is dominated by
the concept of a maximum, game theory has a variety of solution concepts for
predicting the game’s outcomes and the players’ optimal choice of strategies
(e.g., min-max) von Neumann and Morgenstern 1944), Nash (Nash, 1951),
Pareto-optimal (Pareto, 1896), and Stackelberg (von Stackelberg, 1952)).
Maynard Smith, an aeronautical-engineer-turned-biologist, was one of the
first to examine evolution as a mathematical game in his book Evolution and
the Theory of Games (Maynard Smith, 1982). The hawk–dove, sex ratio, Pris-
oners Dilemma, and other interesting matrix games are discussed and analyzed
by Maynard Smith in the context of his concept of an evolutionarily stable
strategy (Maynard Smith and Price, 1973). Our approach to evolutionary game
theory is from a perspective that has its roots in Maynard Smith’s pioneer-
ing work (Vincent, 1985; Brown and Vincent, 1987a,c; Vincent et al., 1993,
1996).
Evolution by natural selection is an evolutionary game in the sense that it
has players, strategies, strategy sets, and payoffs. The players are the individual
organisms. Strategies are heritable phenotypes. A player’s strategy set is the set
of all evolutionarily feasible strategies. Payoffs in the evolutionary game are
expressed in terms of fitness, where fitness is defined as the expected per capita
growth rate for a given strategy and ecological circumstance. The fitness of
an individual directly influences changes in the strategy’s frequency within the
population as that strategy is passed from generation to generation. Evolution,
then, has to do with the survival of a given strategy within a population of
individuals using potentially many different strategies.
Several features distinguish evolutionary games from classical games. First,
the evolutionary game does not fall into one of the three major classes of
games. It is a hybrid with some similarity to both continuous static games and
differential games. Furthermore, in classical game theory, the focus is on the
players who strive to choose strategies that optimize their payoffs; whereas,
in the evolutionary game, the focus is on strategies that will persist through
time. Through births and deaths, the players come and go, but their strategies

pass on from generation to generation. In classical game theory, the players

choose their strategies from a well-defined strategy set given as part of the game
definition. In the evolutionary game, players generally inherit their strategies
and occasionally acquire a novel strategy as a mutation. The strategy set is
determined by genetic, physical, and environmental constraints that may change
with time. In classical game theory, each player may have a separate strategy set
and separate payoffs associated with its strategies. In the evolutionary game,
there will be groups of evolutionarily identical individuals who have the
same strategy set and experience the same expected payoffs from using the
same strategies. In classical game theory, rationality or self-interest provides
the optimizing agent that encourages players to select sensible strategies. In the
evolutionary game, natural selection serves as the agent of optimization. We
know we are dealing with a classical game when admiration is reserved for the
winners (e.g., members of Congress), and we are dealing with an evolutionary
game when admiration is reserved for the survivors (e.g., cockroaches). This is
not to say that winners could not also be survivors, but, as we shall see, survivors
need not be winners in the usual sense.
The evolutionary game has an inner and an outer game (Vincent and Brown,
1984b). The inner game involves only ecological processes and can be consid-
ered as a classical game. For the inner game, players interact with others and
receive payoffs in accordance with their own and others’ strategies. Evolution
takes place in the outer game. It is the dynamical link, via inheritance and fit-
ness, whereby the players’ payoffs become translated into changes in strategy
frequencies.
1.3.2 Games Nature plays

Darwin’s second postulate states that there is a struggle for existence among
organisms. This struggle may be simulated using population dynamics models.
Such models contain many parameters, such as growth rates, resources uptake
rates, predation rates, and carrying capacities. These parameters, in turn, de-
pend on the strategies (i.e., heritable traits) used by the various species in the
population. The influence of strategies on the struggle for existence is obtained
by varying model parameters. If we can identify strategies and embed Darwin’s
first postulate into the system, the strategies in the model will have the capac-
ity to evolve. This gives us all the elements needed for the formulation of an
evolutionary game.
Some elements of the evolutionary game are readily apparent in population
dynamic models. Start by choosing initial population sizes and strategies for
each species in the community. Solve these time-dependent population dynamic

models until the system asymptotically approaches an equilibrium solution (or

until the system develops a persistent state of oscillations or non-equilibrium
dynamics). Those species at a non-zero equilibrium number represent the sur-
vivors. One discovers that, in general, starting with many different strategies
results in relatively few surviving strategies. This represents the struggle for
existence in these models. We seek a special type of stability for the evolution-
ary game that will focus on the main feature: the ability of systems to evolve.
For a given strategy to persist through time, it must be able to maintain a viable
population in the face of the introduction of new strategies and through the
subsequent evolution of the strategies.
1.3.3 ESS concept

It is one thing to be able to characterize natural selection as a game, it is
quite another to determine an appropriate solution concept for that game. Ev-
ery evolving system produces trajectories of changing strategy values. But
do these trajectories have a stable endpoint, and do these endpoints have
anything in common? In particular, do these endpoints of evolution by nat-
ural selection have the common property of producing the best strategy
given the circumstances? One might think that the best strategy is one that
maximizes some measure of collective reward. If this were the case, then
the conventional game concept of a Pareto-optimal solution (Vincent and
Grantham, 1981; Pareto, 1896) would provide an interesting starting point.
However, due to the nature of the evolutionary game, this solution concept is not
appropriate.
Maynard Smith and Price (1973) provided a solution concept for evolution-
ary games with their definition of an evolutionarily stable strategy (ESS).
They reasoned that for a strategy to be evolutionarily stable it must be able to
resist invasion from alternative strategies. With their definition, this resistance to
invasion applied only when almost everyone in the population is using the ESS.
In other words an ESS must be better than all alternative strategies when the
ESS is common throughout the population. Soon it was recognized that the ESS
definition had similarities to the Nash equilibrium (Nash, 1951) of classical
game theory (Auslander et al., 1978). A Nash equilibrium and an ESS are “no-
regret” strategies in the sense that if everyone in a population is playing a Nash
strategy (in a conventional game) or an ESS (in an evolutionary game), then no
one individual can benefit from unilaterally changing their strategy. However,
there are important differences between the Nash definition used in continuous
static games and an ESS. The former focuses on payoffs with no recourse to any
dynamic on the population sizes of individuals possessing particular strategies;

whereas, an ESS must also address population dynamics as influenced by the

strategies present within and among the interacting populations.
The origins of a formal evolutionary game theory can be traced to the in-
troduction of the ESS concept and its application to matrix games. It had its
initial intellectual beachhead on the shores of simple matrix games (Riechert
and Hammerstein, 1983). Matrix games, involving pairwise animal behaviors,
form portions of the conceptual underpinning of almost all current investigations
into animal social behaviors, including behaviors such as territoriality, domi-
nance hierarchies, cooperation, group foraging, vigilance, group size, female
choice, mate competition, and breeding strategies. Evolutionary game theory
has been expanded to include continuous games where strategy sets are contin-
uous rather than discrete (Vincent and Brown, 1984b). Continuous games can
model quantitative traits such as body size, flowering date, and niche specializa-
tion. Because the ESS may contain several coexisting strategies, evolutionary
game theory can be applied to the maintenance of polymorphisms within pop-
ulations (Bishop and Cannings, 1976) or to the question of maintaining species
diversity (Brown and Vincent, 1987a). Evolutionary game theory can apply
to questions of coevolution (Lawlor and Maynard Smith, 1976), speciation
(Dieckmann and Doebeli, 1999), and the evolution of community organization
(Brown and Vincent, 1992). In particular, the last 30 years has seen advances
in the types of ecological and evolutionary dynamics inherent in evolution-
ary game theory, and in the array of stability properties associated with these
dynamics (Metz et al., 1996; Cohen et al., 1999).
Early on it was appreciated that the ESS definition of Maynard Smith cap-
tured one notion of evolutionary stability but missed a second (Taylor and
Jonker, 1978; Zeeman, 1981). The original definition requires only that the ESS
be resistant to invasion by a rare alternative strategy (a necessary condition
for evolutionary stability), but this condition does not ensure that a population
will actually evolve to the ESS. This requires a second notion of evolutionary
stability, that of convergence stability. Convergence stability implies that a
population will evolve to an ESS when its strategy composition is near but not
at the ESS (Eshel, 1983). The presence or absence of these two forms of evo-
lutionary stability yield a variety of possible solutions to evolutionary games.
Notable among non-ESS solutions are evolutionarily stable minima at which
a population’s strategy actually evolves to a fitness minimum (Brown and
Pavlovic, 1992; Abrams et al., 1993b). At such points, an individual by using
its current strategy can do no worse! Remarkably, such perverse situations can
be convergent stable. They are not ESS and interestingly they open the door
to new models of adaptive speciation or raise questions about whether natural
selection actually leaves species with strategies that not only appear unfit, but

seem to be the worst given the circumstances (Brown and Pavlovic, 1992; Cohen
et al., 1999).
1.3.4 Scope of evolutionary game theory

Evolutionary game theory began with the objectives of solving questions of
natural selection involving the behavior of animals confronted with pairwise
conflict. Such conflict is easily modeled by matrix games and the bulk of previ-
ous books on evolutionary games emphasize behaviors and this class of games
(Maynard Smith, 1982; Hofbauer and Sigmund, 1988; Cressman, 2003). Evo-
lutionary game theory can easily model frequency-dependent selection. But,
from a genetics-based perspective on evolution, evolutionary game theory is
often viewed as a short-cut method for solving problems with frequency de-
pendence that applied only to organisms with asexual reproduction. This is a
bit ironic, as sex, genders, and sex ratios are strategies that likely evolved in
response to natural selection and which form a part of the evolutionary game.
Many important issues regarding the application of evolutionary game theory
in the absence of explicit genetics have been resolved. In general, conclusions
obtained using evolutionary game models or their genetic counterparts are the
same or similar. This is particularly the case for quantitative genetics models
and evolutionary game models based upon the adaptive dynamics of quanti-
tative traits. Evolutionary game theory is sufficiently developed to provide a
framework for evolution of adaptations within species, the coevolution of traits
among species, speciation, and micro- versus macroevolution.
Adaptation has been thought of in two contexts. The first of these describes
the FF&F, the extent to which a heritable trait has been shaped via natural
selection within a specified environmental context. The second context is similar
to the first, but not directly tied to the endpoint of natural selection. Adaptation
is often used to describe how a trait or characteristic of an individual serves it
well. For instance, one might refer to how well squirrels have adapted to urban
habitats. This statement conveys how squirrels seem ideally suited to backyards
and how they have used their behavioral flexibility to adjust to peanut butter
for food and attics as homes. However, they have not (at least not completely)
adapted in the sense of evolving new strategies in response to a novel habitat. It
is just that squirrels have a suite of existing traits that permit them to thrive with
humans (this association is most evident with fox squirrels and gray squirrels in
the United States, but this train of thought applies equally well to those specific
mammals and birds that thrive in the backyards of people anywhere in the
world). As a game theoretic concept we define an adaptation as the particular
strategies which make up the ESS. Such a view of adaptation combines the

1.4 Road map 21
appealing parts of the two contexts within which adaptation is commonly used.
The ESS is the endpoint of evolution by natural selection. Furthermore, as a
“no-regret” strategy, an individual’s strategy serves it better than any alternative
strategy. In this sense, adaptations resulting from natural selection are optimal
given the circumstances, and represent the FF&F (Mitchell and Valone, 1990).
We use the term adaptive dynamics (Metz et al., 1996) to describe the
change in the frequency of strategies within the population and the term strat-
egy dynamics to describe how a strategy associated with a particular species
changes with time. Because the term adaptive dynamics has more than one re-
lated meaning in the literature,5 we use it only in the restricted sense above. The
strategy dynamics equations are an important part of evolution by natural selec-
tion that drives a population to an ESS, to other solutions, or to non-equilibrium
evolutionary dynamics. Because of strategy dynamics, Fisher’s Fundamental
Theorem of Natural Selection also applies to evolutionary games.
Co-adaptation in genetics describes reciprocal evolutionary responses of
genes at different loci to each other’s effect on the individual’s fitness. Via co-
adaptation, the gene favored by natural selection at one locus may be influenced
by the genes at other loci. In the context of adaptation, this makes sense when
one considers how different traits interact to determine an organism’s success.
For instance, one often sees co-adaptation between behaviors and morpholo-
gies. Organisms with specialist morphologies often feed selectively, while those
with generalist morphologies feed opportunistically. By allowing strategies to
be a vector of traits, evolutionary game theory can model co-adaptations. In
this case, the ESS for one trait is influenced by the values of other traits within
the organism. For instance, in desert annuals seed-dispersal mechanisms, seed
dormancy, and xeric (dry) adapted leaves can all assist the plant in bet-hedging
against droughts and bad years. Hence, a plant with highly xeric leaves re-
quires less seed dormancy than one with mesic (wet) leaves. Or a plant with
seed-dispersal traits such as hooks for animal dispersal or awns for wind dis-
persal requires less seed dormancy or less xeric leaves. At the ESS we expect
these traits to become co-adapted, and evolutionary game theory provides the
modeling tools needed for such co-adaptations.
1.4 Road map

As with living systems, the evolutionary game theory that we present in this
book evolved from a relatively straightforward theory developed to deal with
5 For example, the term adaptive dynamics also serves to describe changes in gene frequency
from population genetics models. Not surprisingly, the equations for change in strategy
frequency have a parallel to those of Wright (1932) for changes in gene frequency.

biological problems with little structure to a more complex theory that is ap-
plicable to biological problems with lots of structure. Fortunately, the route to
complexity is based on an underlying principle that does not change as we add
layers of complexity. As a consequence, one can understand the basic ideas by
focusing on the easy problem first and leaving the more complicated (and more
interesting) problems for later. We structure this book accordingly.
The bulk of the mathematical development occurs in Chapters 2, 4, 5, 6, and
7. Chapter 2 provides an introduction to population modeling, population dy-
namics, equilibrium points, and the stability of these points. Chapter 3 provides
an introduction to classical and evolutionary game theory. Life is viewed as a
game and the stage is set for game theory to provide the tools for modeling the
ecological and evolutionary dynamics associated with natural selection, with
the introduction of the fitness generating function (G-function) concept. In-
dividuals are said to be of the same G-function if they possess the same set of
evolutionarily feasible strategies and experience the same fitness consequences
of possessing a given strategy within a given environment. There is a close
connection between a G-function and the German term bauplan, which is an
old descriptor for classifying organisms by what appear to be common design
features or design rules. The G-function may be thought of as describing both
an organism’s bauplan and the environmental conditions that the organisms
must deal with.
Chapters 4, 5, 6, and 7 are structured in parallel to address the concepts
needed to model and analyze evolutionary games. In Chapter 4 we present a
recipe for making an evolutionary game, starting with a model of population
ecology and showing how to construct a G-function from it. The G-function
is used in the development of both the ecological and evolutionary dynamics
(Chapter 5). We refer to the combination of population dynamics (ecological
changes in the population sizes) and strategy dynamics (evolutionary changes
in a species’ strategy value) as Darwinian dynamics. The Darwinian dynamics
may converge on population and strategy values that are both convergent stable
and which cannot be invaded by rare alternative strategies. The strategy values
obtained in this way are evolutionarily stable strategies (ESS). Chapter 6
expands on the ESS concept of Maynard Smith to provide a formal definition
of an ESS. An ESS must be convergent stable and optimal in the sense of
maximizing an individual’s fitness given the circumstances and strategies of
others. The ESS maximum principle of Chapter 7 provides necessary conditions
For determining an ESS. In terms of the G-function the ESS is an evolutionary
optimum that represents the FF&F. Our approach in each of Chapters 4–7 is to
present the theory in terms of the simplest problem first before moving to the
more complex problems.

1.4 Road map 23
An ESS can possess a diversity of species or strategies (Chapter 8). These

co-existing species can emerge from both within and between G-functions.
When diversity is promoted within a G-function the evolutionary model can
include speciation and the means for generating diversity. When diversity oc-
curs between G-functions the evolutionary model can include coevolution,
microevolution (evolutionary changes within G-functions), and macroevolu-
tion (evolutionary changes resulting in new G-functions).
Much of evolutionary game theory focuses on matrix games. In Chapter 9
we revisit matrix games from the perspective of a more general theory of evolu-
tionary games. Chapter 10 provides examples for applying the modeling tools
of this book to topics in evolutionary ecology including habitat selection, re-
source competition, plant competition, and foraging games between predators
and prey. The theory also has direct applications to problems involving the
management and conservation of evolving systems (Chapter 11).
In Chapters 4–7 the following classes of biological problems are examined in
detail. The same basic ideas are applicable to each class; however, the features
become more complex as the problems become more complex.
1.4.1 The simplest problem

The simplest problem is one that might mimic early life on Earth. We assume
that the evolving populations are based on a single G-function. There is only
one bauplan and a single (scalar) strategy that defines a phenotype. There is
a corresponding constraint set from which strategies may evolve within upper
and lower bounds (e.g., negative body size not allowed). The adaptive strate-
gies influence the fitness of the organisms relative to one another. Moreover
the abiotic environment is so stable that population dynamics results in stable
ecological equilibria. Evolution in such a system proceeds from the fact that
not all phenotypes can co-exist. Rather, only certain phenotypes or combina-
tions of phenotypes can survive and persist through time when confronted with
a new phenotype. The main objective for the simplest problem and all others
is to be able to identify those phenotypes or species that can persist through
time.
1.4.2 Vector strategies

Even the simplest of organisms possess a number of different heritable traits.
Each functioning protein, each structural character, and each physiological or
behavioral pathway can represent a different, and sometimes independent, heri-
table trait. Many adaptive characteristics of an organism likely influence fitness.

The first generalization of the simplest problem is to introduce vector strategies

with a corresponding constraint set. Many constraints are obvious, while others
are not obvious such as those that result from physical or genetic limitations, or
non-independencies among different heritable traits. Each element of a vector
strategy describes a different heritable trait of interest.
1.4.3 Evolving systems with resources

This class of systems is useful when dealing with plants or animals feeding
on an explicit resource that is not itself evolving. However, each affects the
dynamics of the other. In some situations, competition between the evolving
organisms occurs solely through non-evolving resources.
1.4.4 Multiple G-functions

An important generalization of the simplest problem is the introduction of
additional G-functions. Multiple G-functions are required when a population
contains several bauplans6 or there is a single bauplan with organisms in more
than one environmental setting. This allows the modeling of biological systems
at different tropic levels as well as the introduction of novel types within a given
tropic level. For example, a simple one-tropic-level system could involve just
prey and predators. Two G-functions would be required, one for the prey and
one for the predators. Evolution in this case could result in several prey species
(all of the same bauplan) and several predators (all of the same bauplan). Two
G-functions could also be used to define a simple tropic system with two levels,
for example plants and herbivores. More-complicated systems are studied by
introducing additional G-functions.
1.4.5 Frequency dynamics

The majority of the book takes a population dynamics point of view that deals
with changes in population sizes with time. However, there is an alternative point
of view that is important in the study of genetics and matrix games. Instead of
thinking of a biological system in terms of the density a population has as a
result of its corresponding strategy, one can think in terms of the frequency at
which a given strategy is found in the population. The two points of view are
totally interchangeable; however, there are advantages to both points of view.
6 The original German spelling for the plural of bauplan is Bauplaene. Since we are using this
term in a modern and somewhat different context, we choose to use the English plural.

1.4 Road map 25
The frequency dynamics point of view is particularly useful if there is no density

dependence in the model, often the case in matrix games.
1.4.6 Multistage systems

A given bauplan may be more complicated than a bauplan based on lumping all
life stages together into one variable. While this approach is desirable and useful
for a lot of systems it will not be valid for all. Hence the theory is extended to
include multistage G-functions (e.g., pupa, larva, adult).
1.4.7 Non-equilibrium dynamics

It is generally recognized that equilibrium dynamics represents an idealization,
useful for study but not realistic in the real world. We agree. Having an abiotic
environment that is fixed is not sufficient for stable equilibrium dynamics, but
it certainly helps. Not only does assuming the existence of stable equilibrium
dynamics simplify the theory, but valid conclusions and insights about evolu-
tion can still be made. For those situations where non-equilibrium dynamics is
generic to the system, the theory is extended to include this condition.


01 Understanding Natural Selection

Uploaded by

Copyright:

Available Formats

01 Understanding Natural Selection

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

01 Understanding Natural Selection

Uploaded by

Copyright:

Available Formats

Cambridge Books Online

Evolutionary Game Theory, Natural Selection, and Darwinian Dynamics

Thomas L. Vincent, Joel S. Brown

Book DOI: http://dx.doi.org/10.1017/CBO9780511542633

Online ISBN: 9780511542633

Hardback ISBN: 9780521841702

Paperback ISBN: 9781107406513

1 - Understanding natural selection pp. 1-25

Chapter DOI: http://dx.doi.org/10.1017/CBO9780511542633.002

Cambridge University Press

Cambridge Books Online © Cambridge University Press, 2009

1.1 Natural selection

Cambridge Books Online © Cambridge University Press, 2009

Philosophy encompassed all aspects of the sciences. Then, as today, philosophy

Cambridge Books Online © Cambridge University Press, 2009

Cambridge Books Online © Cambridge University Press, 2009

1.1.2 As Darwin saw it

Cambridge Books Online © Cambridge University Press, 2009

1.1.3 The Modern Synthesis

Cambridge Books Online © Cambridge University Press, 2009

1.2 Genetical approaches to natural selection

Cambridge Books Online © Cambridge University Press, 2009

Viewing evolution as change in gene frequency can produce reasonable re-

Cambridge Books Online © Cambridge University Press, 2009

A consequence of the strictly genetical approach used in current textbooks

Cambridge Books Online © Cambridge University Press, 2009

1.3 Natural selection as an evolutionary game

Cambridge Books Online © Cambridge University Press, 2009

Cambridge Books Online © Cambridge University Press, 2009

Cambridge Books Online © Cambridge University Press, 2009

interference competition. The prisoner’s dilemma characterizes many games

Cambridge Books Online © Cambridge University Press, 2009

Fisher4 in addressing the evolution of sex ratios recognized gender frequen-

Cambridge Books Online © Cambridge University Press, 2009

1.3.1 Game theory and evolution

Cambridge Books Online © Cambridge University Press, 2009

games (Issacs, 1965) are characterized by continuously time-varying strate-

Cambridge Books Online © Cambridge University Press, 2009

pass on from generation to generation. In classical game theory, the players

1.3.2 Games Nature plays

Cambridge Books Online © Cambridge University Press, 2009

models until the system asymptotically approaches an equilibrium solution (or

1.3.3 ESS concept

Cambridge Books Online © Cambridge University Press, 2009

whereas, an ESS must also address population dynamics as inﬂuenced by the

Cambridge Books Online © Cambridge University Press, 2009

1.3.4 Scope of evolutionary game theory

Cambridge Books Online © Cambridge University Press, 2009

1.4 Road map

Cambridge Books Online © Cambridge University Press, 2009

Cambridge Books Online © Cambridge University Press, 2009

An ESS can possess a diversity of species or strategies (Chapter 8). These

1.4.1 The simplest problem

1.4.2 Vector strategies

Cambridge Books Online © Cambridge University Press, 2009

The ﬁrst generalization of the simplest problem is to introduce vector strategies

1.4.3 Evolving systems with resources

1.4.4 Multiple G-functions

1.4.5 Frequency dynamics