Introduction
Adaptive evolution is genetic change in a population due to the correlation
of hereditary variation with variation in reproductive success. Understanding
adaptive evolution involves understanding all aspects of this success -- essentially,
all of ecology and physiology and anatomy and behavior.
The nature of genetic change in populations is understood through the principles
of population genetics and quantitative genetics, as introduced in Chapters
3, 4 and 5.
In these disciplines, the correlation between genetic variation and
fitness variation is simply presumed and is represented abstractly by selection
coefficients (in population genetics ) or by response to selection
(in quantitative genetics). This understanding forms the core of modern
evolutionary theory.
However, neither population genetics nor quantitative genetics explains why
the correlation between hereditary variation and variation in reproductive
success exists in the first place. This correlation emerges from a set
of causal relationships between genetic differences and trait differences
(introduced in Chapter 6, which we have skipped), and a set of causal relationships
between trait differences and fitness differences, which are considered broadly
in the current chapter, Chapter 8. Together these relationships form
a causal loop which yields ongoing adaptive evolution.
Naturalists and biologists have long admired the precision of adaptation.
Understanding adaptive evolution requires us to do more than notice
the "good fit" between each species and its niche. We must
also appreciate how that fit is established and maintained by relative differences
in reproductive success among the individuals of each species. Remember:
Selection cannot favor step-by-step changes toward a trait that might
be beneficial in the future, unless each step offers immediate advantage relative
to alternative variants that are already present in the population.
In answer to the question, "How are particular trait differences related
to fitness differences?", there are as many different answers as there
different species. Unfortunately, determining how particular traits
contribute to fitness in nature is often difficult (see adaptation).
However, in cases where both the fitness costs and the fitness benefits
of a quantitative trait can be measured (or computed), it becomes possible
(in principle) to determine whether adaptation is indeed optimal. A
few general principles can be readily appreciated. Some of these are
introduced in Chapter 8.
Obviously, any adaptive feature of any species might be considered. The
cases which have actually been chosen for intensive investigation are not
necessarily those which seem most interesting but rather those which which
are especially tractable -- for example, those in which all relevant parameters
can be reliably measured and which also permit experimental manipulation.
Optimality (pp. 152-154)
The introduction to Chapter 8 presents the basic concept of optimality.
Many (probably all) significant traits involve an adaptive compromise,
wherein increasing the trait has some benefits but also some costs
and decreasing the trait also has some benefits and some costs. The
presumption is that selection should find an optimum, a trait value
where benefit-to-cost ratio is maximized.
Note the caution at the end of the chapter. The concept of "optimality"
suggests a single best (optimum) value for a trait. However, in some
cases, the "optimum" is not a single best value for every individual
but rather an optimal mix of alternative values (such that what is best for
each individual is frequency-dependent).
Also note that an adaptive optimum is generally a local optimum. The
phrase "local optimum" can be readily understood in terms of an
abstract adaptive landscape, where fitness is represented as altitude.
The global optimum is the highest mountain peak on the landscape.
A local optimum is any bump or hilltop from which movement in
any direction leads downslope. As applied to adaptation, the important
point is that selection always favors variants which have greater local fitness
-- that is, selection always pushes a population uphill on the adaptive
landscape. The result should be eventual arrival at some local peak,
but with no expectation that this peak is the highest point on the landscape.
The "adaptive landscape" metaphor is attributed
to Sewall Wright.
Basic considerations
The next section (pp. 154-156) lists several basic considerations in optimality
modelling.
- Demography. Survival and fecundity are related to age and
size of organisms.
- Heritability. Only heritable variation is susceptible to
optimizing by selection. (Don't be concerned about
the concept of "reaction norms", which was is covered in Chapter
6, p. 118 ff, which we have skipped.)
- Trade-offs. Traits are interrelated both by developmental
genetics and by physiology, so that benefits based on changing one trait
may entail costs through changes in a related trait.
- Phylogeny. All traits are constrained to some extent by history,
by the possibilities and limitations which accompany a particular body plan.
The following points deserve emphasis:
- BOTH benefits AND costs must be considered.
- This consideration must be based on fitness (i.e., relative reproductive
success)
- Costs and benefits must be measured and weighed quantitatively,
across the entire lifespan.
Examples (See text, pp. 156-175, for details and specific examples.)
Optimality modelling is most successful with traits which are related directly
to fitness and for which both costs and benefits are amenable to empirical
measurement. Such traits include most basic demographic parameters:
- Body size.
- Time to maturation (i.e., age at first reproduction).
- Number of offspring produced during a given reproductive cycle.
- Investment (biomass, energy, time) in producing each offspring.
- Total reproductive investment in each reproductive cycle.
- Life span (survival probability as a function of age).
With respect to each of these traits, IF there were no trade-offs:
- Bigger bodies would be better (able to produce more offspring).
- Earlier maturation would be better (shorter generation time).
- More offspring would be better (self-evident).
- Greater investment per offspring would be better (yielding higher
probability of survival).
- Greater investment per reproductive cycle would be better (yielding
more or better offspring).
- Longer life would be better (increasing the number of reproductive
cycles).
However, increasing the benefit from any one of these traits generally involves
a trade-off that decreases the benefit from one or more of the others. Some
of these trade-offs are fairly obvious.
- Body size and maturation time are coupled by growth rate. Earlier
maturation increases the potential reproductive rate and increases the probability
of survival to maturation but also yields smaller bodies which produce fewer
offspring.
- The number of offspring and the investment per offspring are coupled by
available resources. Increasing one decreases the other.
- Reproductive effort per reproductive cycle and life span are coupled by
the probability of surviving to reproduce again. The greater the investment
now, the less reserve is available to promote survival.
In general, the particular balance depends on species-specific ecological
circumstances, so there is no "one-size-fits all" solution to any
optimality problem. That's part of why optimality studies can be so
much fun -- each species poses its own special puzzles.
Extreme examples are provided by the so-called r-strategists and K-strategists.
The labels derive from the equation for population growth, depending
on whether the intrinsic rate of increase (r) or the carrying capacity
of the environment (K) is more relevant to the outcome of selection.
r-selection. In environments where resources
are abundant and survival probability is relatively unaffected by individual
variation, selection will favor profligate production of many cheap offspring
(i.e, lots of tiny eggs).
K-selection. In environments where resources
are limiting and survival probability for each individual depends on its
ability to compete effectively with other conspecifics, selection will favor
exorbitant investment in each individual offspring (i.e., one or two offspring
carefully nurtured to maturity).
Note that much of the text-book presentation in this chapter consists of
extremely abbreviated descriptions rather than careful explanations. The
purpose is to call your attention to a number of curious phenomena where optimality
has been investigated.
Life Span
The evolution of life span is nicely discussed in the text. You
should be acquainted with three different explanations for why genetics should
influence life span (i.e., why evolution does not automatically favor ever-greater
lifespan).
- Ineffective selection against alleles with deleterious
effects that are only expressed after most expected reproductive effort
has already occurred.
- Trade-offs of benefits early in life (when selection is
still effective, before most expected reproductive effort has occurred)
against deleterious effects later in life (when selection is still ineffective,
after most expected reproductive effort has occurred).
- "For the good of the species", to provide opportunity
for the next generation (by eliminating competition from elder generations).
This logical possibility depends on species-level selection and is
difficult to support on theoretical grounds.
Sex Ratio
Be acquainted the following explanations for sex rations.
- Fisher's explanation, based on frequency-dependent selection,
for the common sex ratio of 50:50 (equal numbers at birth of males and females)
- Shaw-Mohler theorem, generalizing Fisher's explanation
to apply to special cases. At the equilibrium sex ratio, parents should
gain the same fitness through sons as through daughters, such that no variant
altering the sex ratio would have a selective advantage. In many circumtances,
the expected equilibrium sex ration will be 50:50.
Additional discussion on the evolution of sex ratios
(separate page).
Conclusion
Optimality modelling can convert "adaptive story-telling" into
quantitative science. It works best as a test of current understanding.
(This is especially true for life-history strategies which are extreme
in one way or another. Extreme strategies are more likely to be based
on unusual circumstances that can be measured.)
If the current value of a trait does not match that predicted
by the optimality model, then one or more of the assumptions that entered
into the model may be unjustified.
Perhaps one has overlooked some significant cost or benefit.
The sign of the difference between predicted optimum and actual trait
value may suggest where to look. (For example, in the case of sexual
reproduction, as discussed in Chapter 7, some substantial
benefit may not yet have been adequately assessed.)
Alternatively, one may have overlooked some constraint,
such as absence of sufficient variability, that prevents attainment of optimality.
In any case, one should avoid using "optimality story-telling"
as a substitute for "adaptive story-telling", in which one simply
presumes that current values for such traits are indeed optimal. One
must also use caution if the optimality model depends to several unmeasured
parameters which may be "fudged" (i.e., adjusted within reasonable
bounds) to compute an optimum which corresponds to the measured trait value.
Notes for chapter 1 / 2
/ 3 / 4 / 5
/ 6 / 7 / 8 / 9
/ 10 / 11 / 12
/ 13 / 14 / 15
/ 16 / 17
Additional considerations.
Trade-offs and optimality represent only one of many themes
for analyzing adaptation. The authors of our text have chosen to devote
a chapter to this theme, but other themes are just as deserving of consideration.
The
theme of arms races is central to coevolution between predators
and prey (or between parasites and hosts).
The "arms race" metaphor, of course, comes from
human military affairs, most notably the build-up of nuclear weapons in
the U.S.A. and the former Soviet Union.
The basic idea is that selection promotes adaptations by
prey (or hosts) to defend against predation (or parasitism), while selection
also promotes adaptations by predators (or parasites) to overcome prey (or
host) defenses. Adaptation by either of the "opposed" parties
constitutes increased selective pressure on the other party to counteract
such adaptation.
The essentially endless quality of the resulting escalation
is captured by Leigh Van Valen's Red Queen Hypothesis,
in which organisms avoid extinction only if they are able to continue evolving.
Symbiosis (or mutualism) is another major theme in evolutionary
adaptation, again involving coevolution, in which organisms
become ever more intensively dependent upon one another. The coevolved
traits are then favorable to all parties.
Familiar examples symbiotic interdependence include the
algae and fungi which comprise lichens, flowering plants and the insects
which pollinate them, and people and their intestinal bacteria.
Many biologists believe that eukaryotic cells represent
ancient symbioses among several once-independent organisms. Mitochondria
and chloroplasts both carry relict DNA that hints at their early, independent
prokaryotic existence.
Relationships established by coevolution are often unstable. In order
to further their own survival and reproduction, parasites can evolve toward
reduced harm to their hosts while hosts can evolve to derive some gain from
the parasite. This can convert a host-parasite relationship into mutualistic
relationship. Conversely, mutualistic relationships can break down into
host-parasite relationships if either party finds a way to increase its own
reproductive success by exploiting the partner without providing any benefit
in return.