Population processes in heterogeneous landscapes are notoriously difficult to study, both empirically and theoretically. There are well-documented effects of scale (both total size and resolution of the system), density dependence, temporal and spatial variation, as well as effects of habitat-shape, -boundary and -edge. One way to study such complex systems is to reduce the dimensions of the system to a minimum of parameters. Thus, the main process studied in this thesis is habitat dependent movement in a heterogeneous landscape. The effect of landscape characteristics (habitat amount and arrangement) on the population processes (population growth and distribution) is modelled using a spatially- and temporally-discrete matrix model framework.

Generally, the contrast in preference to, or habitat specific movement probability in, preferred and secondary habitat was most important to spatial population distribution (Paper I and II). The effect of habitat amount dominated (Paper I) but decreased substantially when habitat arrangement was explicitly included (Paper II). These results differ from other spatially explicit population models, where the (large and negative) influence on population persistence from habitat amount is generally larger than the effect from habitat arrangement This discrepancy can be attributed to the difference in response variables; population persistence is not population distribution. The effect of habitat amount generally increased when habitat dependent growth (both for organisms with one and several reproduction events during a season) was included in the model, but decreased when adding non-habitat dependent dispersal mortality (Paper III). As expected, the amount in preferred habitat (thus correlated with population growth) largely determined other population responses (Paper III). This suggests that the effect of habitat amount and arrangement on natural populations would vary between organisms with differences in within-generation movement behaviour, growth and survival, or at least that results from models with different implementations of movement and demography are expected to differ.

Given the importance of movement, what is a reasonable approximation of within-generation movements? Results from measuring small-scale movements of a Collembola (Protaphorura armata) indicate that food and density of conspecifics have significant effects on movement behaviour (Paper IV). Further, using individual small-scale movements to parameterise a matrix model, without analysing underlying correlation structures, might underestimate dispersal (Paper IV). Thus, reasonable approximations depend on the propensity to disperse, and probably on the shape of the dispersal function.

If within-generation redistribution of a population can be modelled using a matrix landscape model, then there are analyses available to reveal the sensitivity of growth, resilience and population distribution to perturbations of transitions between landscape cells (e.g. through landscape changes; Paper III). I exemplify how landscape managers could analyse the effect on growth and population distribution of the target species in a specific landscape, how specific changes in transitions would affect population growth, as well as what transitions are important for the population density of an area (Paper III).

Applying matrix models to real problems can be difficult at this stage, but matrix models may guide data collection to put efforts where it is most needed. One such conclusion is the importance of non-habitat and the need to correctly estimate how individuals of a population use non-preferred areas.