Landscape connectivity refers to the functional relationship among hab
itat patches, owing to the spatial contagion of habitat and the moveme
nt responses of organisms to landscape structure. Heterogeneous landsc
apes provide a particular challenge for modelling population-level res
ponses to habitat fragmentation, because individuals may be utilizing
multiple habitats to varying degrees across the landscape. We apply ne
utral landscape models to understand how species' habitat affinities i
nteracted with landscape structure (i.e., habitat abundance, distribut
ion, and quality as measured by carrying capacity) to affect the redis
tribution of individuals. Two types of neutral models are presented: r
andom maps, in which the distribution of habitat is spatially independ
ent and fractal maps, in which habitat exhibits an intermediate level
of spatial dependence. The neutral landscapes comprised varying propor
tions of three habitat types, for which species exhibited a preference
gradient (high, medium, low). We performed a series of simulation exp
eriments as a factorial design of parameter states to tease apart the
underlying factors responsible for population distributional patterns
(random vs clumped) in spatially complex mosaics. Landscape connectivi
ty is a threshold phenomenon, in which even a minimal loss of habitat
near the critical threshold (p(c)) is likely to disconnect the landsca
pe, and which may have consequences for population distributions. The
exact value of p(c) depends upon the spatial arrangement of habitat; f
ractal landscapes exhibited connectivity across a greater range of hab
itat abundance (p) than random maps (fractal p(c) = 0.29-0.50, random
p(c) = 0.59). Although the spatial arrangement of habitat (random vs f
ractal) was the most important determinant of population distributiona
l patterns, different landscape factors were important in structuring
populations in the two types of maps. The relative abundance of habita
t had the greatest effect on populations in random landscapes, whereas
scale-dependent patterns were evident in fractal landscapes. At fine
scales, population dispersion was determined by habitat abundance in b
oth random and fractal maps, although populations were more aggregated
(as measured by Morisita's Index, I-m) at this scale in random landsc
apes. But at coarse scales on fractal maps, population distribution wa
s primarily influenced by species' habitat affinities. Assessment of t
he independent effects of habitat affinity and habitat carrying capaci
ty on population distributions revealed that the differential interact
ion of species with landscape structure (i.e., different residence pro
babilities in each habitat type) was the primary determinant of distri
butional patterns. Neutral landscape models thus provide a useful tool
for determining the relative importance of various components of land
scape structure that affect population distributions.