4.3. Theoretical aspects

Patterns in randomly fragmented neutral model landscapes have exemplified the process of habitat loss and its consequences to landscape structure. Independent of the criteria, how model landscapes are subdivided, three main stages in fragmentation process are common. First, structural connectivity breaks down, then the patch sizes decrease and their number increases rapidly, and, finally, the isolation of patches is a direct consequence of reduced patch sizes. Because neutral models eliminate all biological interactions and natural processes in order to produce a random pattern (Caswell 1976), they serve as an important reference background to real landscape studies (With & King 1997). Real landscapes, however, are often naturally fragmented in terms of target habitat and, therefore, e.g. the estimation of critical thresholds in proportion of the original habitat or in habitat isolation are difficult to conclude from the basis of neutral models only. The original pattern and landscape composition in real landscapes should be incorporated in these neutral models as a starting point to determine thresholds for the landscape structure. However, this presumes that the observed landscape structure supports a dynamic population under natural conditions. These conclusions are, on the one hand, species-specific and have to be assessed on the basis of species ecology and plasticity to respond to spatial changes in the landscape structure.

The boreal forest landscape is structurally heterogeneous and from the perspective of the Siberian flying squirrel it is naturally fragmented to some degree. The species inhabits forest sites that are temporary in boreal forest ecosystem (Spies & Franklin 1996) and it encounters habitats that are not suitable for breeding nor dispersing at a local scale (see Tiebout & Anderson 1997). This suggests that the Siberian flying squirrel has adapted to live in a habitat mosaic and use the landscape matrix for colonising novel forest areas. Habitat loss and the amount and extent of totally useless habitat types, however, pose a risk to its local and regional persistence. Habitat loss effect is obvious but the role of pure isolation effect remains relative. If habitats are isolated by wide open areas or sapling stands, isolation may be effective but in case habitat patches are connected by forested habitat (Henein & Merriam 1990, III) even long interpatch distances are traversed and habitat patch networks function as an operative demographic unit.