Stability of resistance/susceptibility and virulence/avirulence polymo
rphisms in a gene-for-gene host-parasite coevolution model was tested
by numerical analysis. Computer simulations were run for 752 different
combinations of parameter values in the model. Repeated simulations w
ith different initial frequencies of resistance and virulence alleles
revealed the presence of an unstable limit cycle for each combination
of parameter values. Represented in a phase plane, unstable limit cycl
es repel gene frequencies (i.e., gene frequencies starting inside the
limit cycle spiral inward toward an internal equilibrium point; those
starting outside the limit cycle spiral outward toward fixation or ext
inction). Depending on their initial frequencies in the model, alleles
for virulence and susceptibility either spiraled toward equilibrium o
r they became fixed. Likewise, alleles for avirulence and resistance m
oved either toward equilibrium or extinction. Thus, the position of th
e unstable limit cycle and the initial gene frequencies determined whe
ther the system went toward a stable equilibrium or fixation of virule
nce and susceptibility. The position of the unstable limit cycle depen
ded on the values of key parameters in the model. For some combination
s of parameter values, the unstable limit cycles extended so far from
the equilibrium point that new genes for virulence could not possibly
enter the parasite population at frequencies outside the limit cycle.
In those cases, the polymorphisms were regarded as stable in biologica
l terms. Two versions of the coevolution model were compared. In the h
ard-selection version, virulence alleles carry an associated fitness c
ost of reduced inherent rate of reproduction on either susceptible or
resistant hosts. In the competition version, only unneccessary virulen
ce carries a fitness cost, because the cost of virulence is expressed
as reduced competitive ability on susceptible hosts. Polymorphisms wer
e stable for moderate costs of unnecessary virulence in the competitio
n version of the model but usually were not stable for the hard-select
ion version. In the competition version, polymorphisms were stable eve
n when there was no cost of resistance, provided that the cost of unne
cessary virulence was moderately high.