I use multilocus genetics to describe assortative mating in a competit
ion model. The intensity of competition between individuals is influen
ced by a quantitative character whose value is determined additively b
y alleles from many loci. With assortative mating based on this charac
ter, frequency- and density-dependent competition can subdivide a popu
lation with an initially unimodal character distribution. The characte
r distribution becomes bimodal, and the subpopulations corresponding t
o the two modes are reproductively separated because mating is assorta
tive. This happens if the resource distribution is unimodal, i.e. even
if selection due to phenotypic carrying capacities is not disruptive.
The results suggest that sympatric speciation due to frequency-depend
ent selection can occur in quite general ecological scenarios if matin
g is assortative. I also discuss the evolution of assortative mating.
Since it induces bimodal phenotype distributions, assortative mating l
eads to a better match of the resources if their distribution is also
bimodal. Moreover, in a population with a bimodal phenotype distributi
on, the average strength of frequency-dependent competition is lower t
han in a unimodal population. Therefore, assortative mating permits hi
gher equilibrium densities than random mating even if the resource dis
tribution is unimodal. Thus, even though it may lead to a less efficie
nt resource use, assortative mating is favoured over random mating bec
ause it reduces frequency-dependent effects of competition.