| The genetic
basis of fitness variation in natural populations
Questions
& Approaches
Our main efforts in the
lab now go towards a project with the annual grass Avena barbata to experimentally
measure the genetic basis of fitness differences between individuals in nature.
Reciprocal transplant experiments usually demonstrate a fitness advantage of local
populations compared to immigrants transferred in from other locations. This implies
first that genetic variation has consequences for survival and reproduction, and
second, that some form of constraint prevents populations from becoming well adapted
to all environments, even though heterogeneous environments would seem to favour
such broad adaptation
How many genes are involved
in adaptation to the local environment, and are these the same genes across environments?
One hypothesis posits a trade-off with the same genes are under selection
for alternate alleles in contrasting environments. In this way, recombination
could not create a broadly adapted genotype, and heterogeneous environments would
maintain genetic diversity. An alternative hypothesis suggests that local adaptation
may arise through the accumulation of mutations that are neutral in the native
environment, but detrimental in novel environments. Such mechanisms do not promote
the maintenance of genetic variation and implies that different genes are involved
in adaptation, and that novel combination of alleles at these genes should be
well adapted in both environments.
An important related question is:
What is the evolutionary fate of hybrids between divergent
populations? Can recombination produce broadly adapted genotypes? Alternatively,
do specific combinations of alleles across loci interact epistatically to create
the adaptations to particular environments? If epistasis contributes to local
adaptation, does the creation of novel combinations across loci permit the adoption
of novel ecological niches?
We are integrating two approaches to answer
these questions. The classic biometric approach
partitions the phenotypic variation into genetic and environmental components;
measures the phenotypic selection gradients acting on traits; and estimates correlations,
tradeoffs and genotype by environment (GxE) interactions. This approach thus provides
information on the variation that selection "sees" and can act on at any given
time. More recently, quantitative trait locus (QTL) mapping
has made it possible to identify specific chromosomal regions which underlie
phenotypic variation. These allow us to identify “genes for” traits, and to ask
specific questions about the action of these genes. This approach thus provides
information on the underlying genetic mechanisms creating phenotypic and fitness
variation
Study
System: Avena barbata (Slender Wild Oat)
The Slender wild oat Avena barbata (Pott ex Link, Poaceae, Picture
Right) is a selfing annual grass, that shows two distinct ecotypes associated
with moist vs dry natural environments in California (Allard, 1999). Moreover,
because it is also an obligate annual, lifetime reproductive success (ie, fitness)
is easily measured as the number of seeds produced before senescence. Each spikelet
produces two single-seeded florets, and the glumes of the spikelets are retained
on the plant after seeds drop so that an accurate count of seed number is possible
by simply counting spikelets and multiplying by two. Avena is easy to
raise in the greenhouse or field, and we have permanent
plots set up at the Hopland
(moist) and
Sierra Foothill (dry) Research and Extension Centres operated by the University
of California.
I have crossed the two ecotypes, producing 188 Recombinant Inbred Lines (RIL's
Figure left). These RILs have each been propagated to the F6 stage by single-seed
descent, which eliminates as much as possible the opportunity for selection to
act during this propagation. Thus, the RIL's represent a set of completely homozygous
lineages each fixed for a random combination of alleles from the two parents.
Because it is naturally selfing, Avena tolerates this inbreeding well
– indeed, it is exactly what would occur naturally after a cross in the wild.
The use of RILs increases the efficiency of the study greatly. Because
unlinked genes have been shuffled by recombination, the only markers still showing
linkage disequilibrium are those that are adjacent to one another on the chromosome,
and this allows efficient linkage mapping of the genome. Similarly, any correlation
among traits implies that the traits have a shared genetic basis and thus a potential
constraint on adaptation. Finally, an association between traits and markers in
this mapping population implies that there is a QTL that affects the trait located
somewhere near the marker
Results so Far
Genetic Correlations : Functionally related traits seem to be strongly
correlated. In the greenhouse, the main target of selection seems to be flowering
time. Kyle Gardner has shown a very strong (r = 0.55-0.6)
genetic correlation between flowering time, reproductive allocation and fitness
– those families that flower early, allocate more of their resources to reproduction
(and less to vegetative growth) and produce more seeds than those that flower
late. In the field, fitness is most strongly correlated with mass
But
other traits show much les correlation. Joanna MacKenzie
and Angel Vats have measured the competitive
ability and root allocation of the parents and recombinants. While the Mesic Genotype
is a better competitor and the Xeric has higher root allocation, these two traits
do not show any correlation among the recombinant progeny (
Figure ). So it appears that recombination can produce novel combinations
of ecological abilities increasing variation with which to adapt to new ecological
niches
Most importantly, the performance of RILs across environments
shows a weak, positive correlation, rather than the negative correlation one would
expect if there were a tradeoff between fitness in different environments. Thus
recombination has indeed produced genotypes that have high fitness in multiple
environments
Mapping :
Kyle Gardner generated the genetic map of our RILs using 130
AFLP markers . The markers show good Mendelian segregation (Figure) and resolved
19 linkage groups spanning 644 cM. Kyle found QTL for all traits, explaining from
5 to 50% of the variation among lines. The salient results are:
The
traits that are strongly correlated (flowering time, reproductive allocation and
fitness) trace to the same QTLs. That is, correlated traits have a common genetic
basis
Fitness in different environments maps to different QTL. Recombining
these QTL produces the broadly adapted recombinants. At no locus did we find selection
favouring alternate alleles in different environments
QTL alleles are
linked in repulsion phase in the parents ( Figure
). That is, at some loci, the selectively advantageous allele is in the Mesic
parent, and at others, the advantageous allele is in the xeric parent. In these
cases, recombining the alleles can produce genotypes with high fitness alleles
at all loci, and these should have higher fitness than the parents
Fate of Hybrids :
Joey Johansen applied line cross analysis to measuring the relative fitness
of early (F2) vs late (F6) generation hybrids across a range of environments.
Late generation hybrids are somewhat less fit than the parents indicating a disruption
of coadapted gene complexes. BUT
(a) in early generation hybrids, where
many loci are still heterozygous, the hybrid vigour brought about by this heterozygosity
seems to counteract the breakup of coadapted complexes and
(b) while the average recombinant has lower fitness
than the average of the parents, certain individual
recombinants are more fit than either parent. The lines that outperform
the parents contain all the favoured alleles at the QTL Kyle found affecting greenhouse
fitness.
Even more intriguing, this fitness advantage seems to hold
across a range of environmental treatments in the greenhouse – could these be
broadly adapted genotypes? But in the field, their advantage is less pronounced
while hybrid breakdown is more pronounced. Since the greenhouse is a novel environment
(compared to the field), it would seem that recombination is most beneficial when
adapting to a novel environment
Genotype by Environment Interactions and Phenotypic Plasticity: One
of the interesting findings to come out of Joanna MacKenzie's work was that the
recombinant lineages show a significant heritable variation for Phenotypic Plasticity.
Most plants exhibit some ability to alter their morphology in response to the
environment - a common example is to increase root growth indrier conditions.
What Joanna found was that the recombinant lines differed in their ability to
do this. Katy Schurman followed up on this finding
to ask: Does plasticity serve a benefit to the plant in allowing it to perform
well across a range of environments? If so, does this benefit come at the cost
of weakened performance in single environments. Katy measured the degree of plasticity
for several traits and correlated this with mean fitness across environments.
For most traits she found no cost to plasticity - those genotypes which changed
the most did not do any worse in individual environments. Surprisingly however,
she found that some of the more plastic genotypes actually did worse - possibly
it is homeostasis that is favoured for these traits. |
the
wild oats project
(C) Carol Whitham 
|
