Within a population there are frequently several discrete morphs. Whil
e in some cases, particularly color polymorphisms, this variation can
be explained by simple Mendelian modes of inheritance, in many Gases t
he evidence suggests a polygenic pattern of inheritance. The threshold
model of quantitative genetics, in which discrete morphs are determin
ed by some underlying continuously distributed trail and a threshold(s
) of expression, is applied appropriately in these cares. The discrete
morphs exhibited in cyclomorphosis, pedomorphosis, pedogenesis, ''pro
tective'' dimorphisms, trophic dimorphisms, wing dimorphism, and matin
g strategies can all be analysed by using this model. Analyses of a wi
de range of different types of threshold traits show that there is typ
ically a large additive genetic component, but that there is also stro
ng environmental induction. A review of studies shows that no morph ha
s a universally higher fitness, but that there is a tradeoff with the
relative fitnesses of two morphs being contingent upon environmental c
onditions. For example, exuberant structures that serve to protect org
anisms from predators reduce other components of fitness, such as deve
lopment time and fecundity. Environmental induction is an adaptive non
of reaction, in that cues of current or future conditions are used to
increase the Probability that the morph produced is that which has th
e highest fitness under the expected conditions. Most models for the e
volution of threshold traits have focused on the phenotype and have no
t addressed the crucial question of what maintains genetic variation,
and hence permits continued evolutionary change. Phenotypic models sho
w that noninducible polymorphic variation cannot be maintained by spat
ial variation alone, but can be favored in an environment that is temp
orally variable. Multiple phenotypes may evolve in a spatially variabl
e environment if there are cues that allow the organism to assess the
type of patch in which it is developing. thus spatial variation is exp
ected to lead to the evolution of inducible phenotypes. Considerable g
enetic variation can be maintained by mutation, even in the face of st
rong directional selection. Frequency-dependent selection, shown to pl
ay an important role in the maintenance of phenotypic variation, may a
lso be significant in the maintenance of genetic variation.