deleterious effects that may accumulate over years at low
densities, and occasionally will produce genetic novelties
that will be given a chance to test their fitness potentials. A
particularly instructive example of this life style is that of
Rattus villosissimus,which lives in semiarid and arid Aus-
tralia (Newsome and Corbett 1975). This species persists in
refugia that result from unpredictable local rainfall events.
The locations of such refugia are therefore unpredictable.
Occasionally, the desert experiences a more general precip-
itation episode, which allows the rat to increase greatly in
numbers, disperse widely, followed by restriction back to
survival in a new set of small refugia.
Spatial Structure and Movements
The spatial structure (dispersion) of a species is closely re-
lated to its demographic behavior, which, as we have seen,
signals its propensity to suffer extinction and also gives
us guidelines for appropriate conservation actions. Spatial
structure also strongly influences genetic structure, which
can feed back to demographic processes (Chepko-Sade and
Halpin 1987; Lidicker and Patton 1987). Clearly this spa-
tial structure /demography connection is something we need
to understand for all species of conservation concern. A spe-
cies’ distribution in space depends on its geographic distri-
bution, fragmentation of habitat, dispersal capabilities, and
social system. Two of these factors sustain strong anthro-
pogenic influences. Geographic range is affected by habitat
destruction and global climate changes, the latter includ-
ing temperature, the amount, seasonality, and regularity of
precipitation, changes in sea level, and frequency of severe
storms. Habitat destruction clearly causes the fragmenta-
tion of destroyed habitats and the coalescence and avail-
ability of new habitats. The other two variables are less af-
fected by humans as they are inherent species properties.
The social system can interact with genetic structure in var-
ious ways (Chepko-Sade and Halpin 1987; Lidicker and
Patton 1987). The proportion and sex ratio of young that
remain philopatric or that disperse to various distances can
be strongly influenced by social factors (Anderson 1989;
Wolff 1999). At high densities, dispersal can be inhibited
by a tight mosaic of hostile territorial defenders, producing
the so-called “social fence effect” (Lidicker 1976; Hestbeck
1982; Wolff 1999).
Increasing fragmentation of habitats means that more
and more species are living in a metapopulation structure,
that is, as an array of variously isolated populations (demes),
variously connected by dispersal (McCullough 1996; Han-
ski and Gilpin 1997; Krohne 1997; Hanski 1999). The suc-
cess of a metapopulation depends on the balance between
the chances of extinction in the habitat patches (demic mor-
tality rate) and the chances of colonization of empty patches
by a sufficient number of individuals for establishing a vi-
able population (demic reproductive rate). In general, large
and high-quality habitat patches suffer less demic extinc-
tion than small and low-quality patches. But this is not in-
variably the case, because large patches may permit the res-
idency of predators that would be absent or transient in
small patches, and hence result in reduced numbers or even
demic extinctions (Oksanen and Schneider 1995; Lidicker
1999). Moreover, large and dense demes risk decimation by
parasites or other pathogens.
The rate of colonization is a function of successful dis-
persal among patches. This in turn is influenced by a spe-
cies’ inherent vagility, the timing and magnitude of emi-
gration from patches, the nature of the matrix habitat, the
presence of dispersal barriers, a species’ response to habitat
edges and corridors, mortality rates during dispersal, dis-
tances among patches, and the social system (Wolff 2003b).
With all of these variables at play, it is notoriously difficult
to predict successful colonization rates for any particular
metapopulation. Moreover, empirical measures of coloni-
zation rates are often difficult and may require long-term
studies.
One feature of organisms that can impact a number of
these relevant variables is the presence or absence of pre-
saturation dispersal(Lidicker 1975; Stenseth and Lidicker
1992). This type of dispersal behavior generates emigrants
even during the growth phase of populations. Moreover,
these emigrants are in relatively good condition and gener-
ally travel during favorable times of the year. Both the qual-
ity and quantity of emigrants are thus enhanced by pre-
saturation dispersal, and species exhibiting this behavior
are likely to have a relatively high colonization rate. A sec-
ond critical variable is the species’ response to habitat edges
(Lidicker and Koenig 1996; Lidicker 1999; Lidicker and Pe-
terson 1999). Species that avoid edges or that are reluctant
to cross into matrix habitats will be unlikely to locate cor-
ridors between patches and even more unlikely to search
for distant habitat patches. In his study of forest /clearcut
edges in Oregon, Mills (1995, 1996) illustrates a variety of
edge responses by rodents. Chipmunks (Tamias townsendii)
avoided clearcuts but not the forest edge; red-backed voles
(Clethrionomys californicus) avoided both edges (to a dis-
tance of 45 m into the forest) and clearcuts; and deermice
(Peromyscus maniculatus) preferred clearcuts but extended
into the forest in diminishing numbers to a distance of 45 m.
A nonrodent (Sorex trowbridgii) did not seem to recog-
nize the boundary as an edge. In patches of grassland main-
tained by mowing the matrix, Microtus pennsylvanicusfa-
vored patch edges, presumably because it could feed on
freshly sprouted vegetation in the adjacent matrix (Bowers
and Dooley 1999), whereas Microtus canicaudusfemalesIssues in Rodent Conservation 457