cuprina. Initially, resistant genotypes were at a
selective disadvantage relative to the suscepti-
ble genotypes in a diazinon-free environment.
But, after the resistance became widespread,
resistant genotypes were no longer at a selec-
tive disadvantage in the absence of the insec-
ticide when they also contained a particular
mutation located at a different chromosome.
The Role of Population Size
Population size plays an important role in the
adaptive recovery process, in at least two
ways. First, a sharp decrease (bottleneck) in
population size will enhance the possibility of
fixation of a deleterious mutation in the popu-
lation by genetic drift. This would provide the
starting-point of subsequent compensatory
evolution. Secondly, the population size dur-
ing the recovery process affects the probabil-
ity of various types of fitness-restoring
mutations. Basically, the available evidence
suggests that instantaneous and complete
recovery can only be achieved by very rare
mutational events, such as the precise back-
mutation that reverts the genotype to its origi-
nal state. On the other hand, many different
mutations are often possible that restore fit-
ness to a lesser degree. In a very large popula-
tion, the single unique back-mutation might
perhaps be expected to occur within a reason-
ably short time period and its large selective
advantage might cause its rapid spread. But,
in smaller populations, this mutation is
unlikely to occur and selection will promote
the spread of the more common compen-
satory mutations of small effect.
Several experiments have shown this latter
scenario to be realistic. Burch and Chao (1999)
subjected the bacteriophage 6 to intensified
genetic drift and caused viral fitness to
decline following the fixation of a deleterious
mutation. They then propagated the mutated
virus at a range of population sizes and
allowed fitness to recover by natural selec-
tion. Typically, it was recovered in small steps.
Step size during recovery was smaller with
decreasing size of the recovery population.
This result suggests that mutations improving
fitness by a small amount are more common
than those with bigger positive effects. Burch
and Chao also demonstrated that the advan-
tageous mutations of small effect were com-
pensatory mutations whose advantage is
conditional on the presence of the deleterious
mutation that caused the fitness decline.
Levin et al.(2000) performed similar experi-
ments using streptomycin-resistant mutants
of Escherichia coli. They showed that the fit-
ness recovery is mediated primarily by inter-
mediate-fitness compensatory mutations,
rather than by high-fitness revertants, and
that this result is dependent on the numerical
bottlenecks associated with serial passage in
their experiments.Conclusions
The evidence discussed above suggests that
recovery following a fitness reduction is often
of a compensatory nature, in particular if pop-
ulations are experiencing occasional bottle-
necks in numbers. This means that fitness is
restored not by removing the deleterious gene
or genotype, but by the spread of mutations
with a beneficial effect only in the presence of
the gene that caused the fitness reduction.
This gene or genotype will therefore remain
(for longer) in the population. An example is
provided by mutations causing resistance to
antibiotics or pesticides that remain present
after the application of the relevant toxins has
been terminated. However, the principle may
apply to other deleterious mutations as well,
such as a virulence-reducing mutation in an
agent used in biological control. Were such a
mutation to be fixed in a population due to
the passage through an extreme numerical
bottleneck, fitness might be restored by com-
pensating mutations that do not affect viru-
lence directly and the reduction of virulence
might thus become a trait that would be diffi-
cult to remove from the population.
What might the meaning of these findings
be for mass production of biological control
agents? If the aim is to rear natural enemies
that are similar to the initial field-collected
population, one should either prevent bottle-
necks and always keep large populations or
one should replace laboratory populations
that have experienced a bottleneck with new,
field-collected material.Adaptive Recovery after Fitness Reduction 91