The Red Queen: parasitism, arms races, and the origin and
maintenance of sexual reproduction
Ben Bolker
Parasitism and infectious diseases lead to ...
antagonistic coevolution between hosts and parasites, or
arms races. As it turns out, this antagonistic coevolution
is one potential solution to another evolutionary paradox ...
1 The cost(s) of sex
One of the things we tend to miss as a result of our own, human
biology and culture is that in a perfect, homogeneous world where we
were just trying to produce as many copies of our genes as possible,
as quickly as possible (which is after all the name of the Darwinian
game), sex would be a really bad idea. (This is one of the greatest
strengths of theory in biology-making us see that there
are great questions about some of the things we don't see in
nature, rather than just explaining the things we do
see.)
There are several theoretical problems with sex, often referred
to as costs of sex.
Cost of outbreeding: too much variation could be bad, if
sexual recombination breaks up co-adapted gene complexes. This would
lead to what is called outbreeding depression; hybrid sterility
or inviability is the extreme case of this, when an organism mates
across species lines.
Cost of mating: in environments where population
densities are very low, it can be hard to find a mate.
Activities associated with mating can be costly
(in terms of time or energy) or risky
(in terms of predation or transmission of disease).
Cost of male function: why spend time, energy and nutrients
producing males when you could be producing females? (Parthenogenetic
females manage quite happily producing females alone.) Even if you
are a hermaphrodite, it still costs more to maintain the machinery to
produce both sperm and eggs than to produce eggs alone.
Cost of meiosis: why pass on only one copy of your genes when
you could pass on two? This is often considered the strongest and
most basic cost of sex, because it imposes a 50to asexual lineages: somehow, sexuals have to be twice as fit to make
up for the cost of meiosis.
There are many variants on sexuality, ranging all the way from perfect
clonal reproduction through various degrees of mixing (ploidy cycles;
selfing hermaphrodites [with variants among plants like
monoecy, having male and female flowers on the same plant and
monocliny, having pistils and stamens within the same flower];
outcrossing hermaphrodites; and dioecy, having separate male and
female organisms. It's worth thinking a bit about whether all
of the aforementioned costs apply to all of these examples ...
the evolution of sex is also closely tied to the evolution of
recombination.
The cost of meiosis is the largest and most general of
these costs, and applies generally across all sexually reproducing
organisms; it implies that sexual organisms ought to have at least a
twofold advantage, somehow, in order to maintain themselves. (The only
counterexample would be if "all else were not equal", and sexual
individuals could produce twice as many offspring as asexuals [to
produce the same net number of gene copies] - but this is typically
not true.) [See model in Lively 1996.]
2 Hypotheses for the maintenance of sex
So, in that case, what benefits might sex have to allow it to have
started in the first place/be maintained over evolutionary time?
- obligate sex Some taxa are more
flexible than others. For whatever biochemical
or cytological reasons, mammals and birds just
don't seem to be able to switch to asexual reproduction.
Herps and fish are more flexible, inverts do it
a lot, plants do too. Constraints, of some variety, on their
reproductive biology are the proximate
reason why mammals don't reproduce asexually. This still
leaves open the question of why their ancestors would
go down this one-way street.
- Kondrashov's categories:
- Mutation stochastic:
population genetic mechanism based
on higher mutation load in asexual
populations, driven by randomness/genetic
drift/small populations:
Muller's ratchet.
Given what we know about mutation load, probably
only works for small (10-100 individual) populations
- Mutation deterministic:
if there is negative epistasis - deleterious
mutations have compounding effects - then
sexual populations actually do better
(counterintuitively) because recombination brings
deleterious mutations together and allows them to
be more easily purged from the population.
This mechanism works best in large populations,
and does not rely on genetic drift. It's been
called (not by Kondrashov) Kondrashov's
hatchet. The evidence for it (population genetic
parameters measured in natural populations)
is equivocal.
- Environmental stochastic: ecological
mechanism based on the supposed higher variability
of sexual populations/offspring. If the environment
varies over time in such a way that some genotypes
have very low (or zero) fitness in some years
(hard selection), independent of the size of
the population,
then producing a wide variety of offspring - at
least one of which is able to survive in the
bad years - is advantageous (long-term fitness
goes as the geometric average of fitness
over time). Lively calls this the lottery
model.
A second variant on this is the tangled bank
model, which says that (spatial, not
temporal) environmental heterogeneity interacts with
competition. Rare individuals do best, because they
can occupy uncrowded niches. Sexuals do better because
their offspring are more variable, and hence have
a better chance of being rare.
- Environmental deterministic:
a way to maintain rare advantage/advantage of
variability without relying on stochasticity in the
environment is to depend on the existence of
parasites which multiply to target the most
common genotypes (Red Queen hypothesis).
In all of this, we need to be careful distinguishing the
true effects of sexual reproduction. Ecologists tend to assume
it produces "more variable" offspring, but this is not
necessarily the case. What sex really does is
to allow recombination of different genotypes
... what is the true relationship between sexual reproduction and
variability? It depends on population size, how frequently asexual
lineages are split off from the sexual population and how, etc. etc..
How can we test these different hypotheses?
It's tricky.
As with many other evolutionary questions, which are too deeply
rooted to mess with experimentally (i.e. we can't make sexual and
asexual versions of organisms experimentally), we have the
following kinds of options:
Correlational studies: where would we expect to see
each one? Reproductive assurance = low population densities;
lottery model = high-variability environments; tangled bank =
high-density environments; Red Queen = high-parasite environments
Observational studies: are the basic (qualitative) assumptions of
the models satisfied? What about the quantitative requirements for
them to operate in real life? What about corollaries/auxiliary
predictions of each hypothesis?
Detailed models:
- gene-for-gene: avirulent and virulent parasite
alleles; susceptible and resistant alleles. Susc. hosts can
be infected by any parasite. Resistant hosts can only
be infected by virulent parasites. Multiple loci ®
host can resist if it blocks at any allele
| A1B1 | A1B2 | A2B1 | A2B2 |
| A1B1 | + | - | - | - |
| A1B2 | + | + | - | - |
| A2B1 | + | - | + | - |
| A2B2 | + | + | + | +
|
In this example, the host genotypes are the columns and the
parasite genotypes are the rows: A and B are two loci, each
with two alleles. Allele 1 is a susceptibility allele in the
host and an avirulence allele in the parasite. Allele 2 is
a resistance allele in the host and a virulence allele in the
parasite. For example, parasite genotype A1B1 (which
has no virulence alleles) can only infect host genotype A1B1
(which has no resistance alleles); parasite genotype A2B1
can infect hosts A1B1 and A2B1, because it can overcome
the host resistance allele A2, but it can't infect hosts
A1B2 and A2B2 without virulence allele B2.
In gene-for-gene systems, virulence and resistance alleles
tend to accumulate. Once there is a resistance allele (e.g. A2)
in the plant population, it tends to spread (because resistant
plants have an advantage in the presence of avirulent parasites);
it drives the spread of the virulence allele in the parasite
population (because virulent parasites have an advantage in
the presence of resistant hosts). Once the parasite virulence
allele has spread through the population, however, the
resistance allele in the host is no longer of much use. If
there is a cost of resistance, then the resistance allele will
then decline in frequency. If there is a cost of virulence,
then the virulence allele will decline once the resistance
allele has declined.
- matching alleles: parasite match exactly
to infect
| A1B1 | A1B2 | A2B1 | A2B2 |
| A1B1 | + | - | - | - |
| A1B2 | - | + | - | - |
| A2B1 | - | - | + | - |
| A2B2 | - | - | - | +
|
Matching-allele systems lead more naturally
to cycling. Parasites that match a common
host genotype will do better, leading to the
depression of that host genotype, the rise
of other host genotypes, and the subsequent
rise of other parasite genotypes.
© 2005 Ben Bolker
File translated from
TEX
by
TTH,
version 3.66.
On 10 Jan 2005, 13:26.