Mixing and asynchrony of interactions can be expected to stabilize the
dynamics of populations. One way such mixing occurs is by dispersal,
and Hastings and Gyllenberg et al. have shown that symmetric dispersal
between two local populations governed by logistic difference equatio
ns can simplify the dynamics. These results are extended here by using
a more flexible difference equation and allowing asymmetric dispersal
. Although there are some instances where dispersal is destabilizing,
its stabilizing effect is enhanced by asymmetry. In addition, very hig
h dispersal rates can induce a stable equilibrium of the metapopulatio
n despite highly chaotic local dynamics. If this equilibrium loses sta
bility, the route to intermittent chaos can be observed. Two new condi
tions under which dispersal can be stabilizing are discussed. One occu
rs when the timing of reproduction and dispersal differs in the two pa
tches of the metapopulation. This enlarges the asynchrony of the inter
actions, and simple dynamics due to dispersal are more likely. The sec
ond works by slightly adjusting dispersal rates to control chaotic dyn
amics. The control can replace chaos by a stable equilibrium. The evol
ution of dispersal rates is discussed. Since obtaining general criteri
a for invasion into a population with chaotic dynamics is difficult, n
o clear conclusions are possible as to whether evolution leads to more
stable metapopulations. However, a mutant that controls chaos can inv
ade a resident having the same local dynamics but no control mechanism
, so that evolution can lead from chaos to a stable equilibrium. (C) 1
995 Academic Press, Inc.