4.1.1. Spruce dominated forest patch: a basic landscape element for the Siberian flying squirrel in northern Finland

Boreal taiga landscapes are diverse and heterogeneous in habitat types and their spatial and temporal patterns. Despite this general heterogeneity, forest habitat patches in landscapes are dominated by few tree species. In Fennoscandian, boreal forests are mainly dominated by spruce and pine while broad-leaved trees occur principally on more nutrient rich soil types. Deciduous trees are also abundant during early stages of forest succession or in southern parts of boreal forest vegetation zone. The Siberian flying squirrel inhabits spruce and spruce dominated forest habitats in Finnish boreal taiga and, thus, the grain of the landscape for the species is delineated by individual spruce dominated forest patches (sensu Wiens 1989, II, IV, V). Although the species seems to have accepted rural and semiurban habitats in southern Finland, the extent of its distribution in northern Finland is determined by the amount of spruce forest and its spatial dispersion (Wiens 1989, I, III). At the habitat patch scale, the Siberian flying squirrel responds to the internal structure of a spruce forest patch (IV, V). The occurrence of the species in habitat patches has been shown to be associated with deciduous trees (Eronen 1996, Hanski 1998). Deciduous trees provide food and cavities for safe nesting and roosting, resources considered necessary for the species (Hanski et al. 2000). Deciduous forest sites are patchily located in natural forests and they are relatively small in size in the north of Finland. The size of a spruce forest patch and the amount of deciduous habitat inside the spruce forest determine the quality of the habitat patch, and affect the probability that the habitat patch is occupied in northern Finland (V) (Table 1).

But are spruce dominated forest habitat patches the smallest landscape units to which the species is responding? Radiotracking studies have shown that adult individuals move among habitat patches and males especially visit even remote habitat patches frequently (Hanski et al. 2000, own observations). This suggests that a single habitat patch as such might not be a sufficient landscape unit for individuals in landscapes where habitat patches are in general small in size but instead a group of habitat patches forms an ecologically functional unit for persistence. Individuals may complement and/or supplement (sensu Dunning et al. 1992) necessary resources (nutrition, nests, mates etc.) that are spatially and temporally dispersed among habitat patches.

4.1.2. Landscape connectivity at local scale

In order to estimate landscape connectivity, landscape structural characteristics that promote the movement of a focal species in a landscape matrix need be identified. Also information on functional characteristics such as immigration and emigration rates from habitat patches and survival probabilities of individuals in a matrix during movements should be available (Tischendorf & Fahrig 2000a, b). Because these data are seldom obtainable, landscape connectivity in empirical studies refers in practice to structural characteristics in landscapes. The lack of demographic data on the Siberian flying squirrel does not allow to fully incorporate the functional aspect of connectivity into landscape connectivity measures.

At a local scale, landscapes that contain occupied habitat patches are structured in a coarse grain manner and they are less fragmented i.e. there are fewer but larger habitat patches for movement than in northern Finnish forest landscapes on average (c.f. Rolstad & Wegge 1987). Coarse graininess enhances landscape connectivity by providing larger uniform landscape units for movement. Habitat edges and the increase in contrast between habitat patches is likely to lower the probability of crossing the boundary (Wiens et al. 1985, Åberg et al. 1995). Sharp edges will block the movement of the Siberian flying squirrel if moving by gliding from tree to tree is prevented. However, narrow gaps are not likely to affect the use of space of the species if open areas can be crossed by gliding (own observations). The effective landscape use at the local scale is enhanced if breeding habitat patches are embedded in the matrix of habitat where interpatch distances can be traversed along more or less forested habitat (V). Thus, at this scale, landscape connectivity for the Siberian flying squirrel is principally a matrix effect and the quality of matrix determines the degree of connectivity in a landscape (Table 1). However, individuals are unlikely to move along the shortest distance between two habitat patches but they tend to move along closed canopy forest habitat and cross narrow gaps. Therefore, interpatch distances should not be measured by using shortest Euclidean distance but by using landscape structural characteristics as a clue of landscape connectivity. This, however, provides good ecological knowledge on the species ecology. The dispersal potential of the Siberian flying squirrel ranges up to nine kilometres (Selonen & Hanski 2000) but some individuals are likely to disperse even further. The maximum dispersal distances normally cover several small and a few large habitat patches in northern Finnish forest landscapes where the species is rather abundant. Within the average dispersal distance of two kilometres there are always suitable habitat patches for the Siberian flying squirrel. Nevertheless, the role and proportion of long distance dispersers in population dynamics is not known (see Fahrig & Paloheimo 1988).

4.1.3. Landscape configuration of habitat patches

Habitat patch configuration has been considered an important structural characteristics in a landscape but for mammals and arboreal species, in particular, landscape configuration is linked further to landscape connectivity. For the Siberian flying squirrel interpatch distances of occupied habitat patches tended to be shorter than what was observed for unoccupied ones (V). The proportion of open areas surrounding large old-growth remnants was connected to lower probabilities in occupancy (I). However, the composition of the surrounding landscape at local scales seemed to be rather similar for occupied and unoccupied habitat patches and had no significant effect on their occupancy probabilities (V). This suggests that the effect of physical arrangement of habitat patches is masked by other landscape characteristics or classification criteria of landscape elements. This may also indicate that landscape configuration was perhaps measured at the wrong spatial scale. At a local scale, structural connectivity and the fact that most of the patches are within the dispersal range of the Siberian flying squirrel are far more important than configuration (Table 1). Regional scale studies (I, III) did not detect any patterns in the intervening landscape which is most likely a result of too coarse a classification or too large a spatial scale. Moreover, there are no unambiguous and satisfactory quantitative measures to characterise the dispersion and juxtaposition of habitat patches in a spatial context. Indices that describe spatial dispersion of landscape elements have often been developed in model landscapes and are therefore difficult to interpret in real landscape situations (Schumaker 1996, Hargis et al. 1998). Quantification of spatial arrangement of landscape elements would be important to illustrate landscape configuration more accurately. In boreal forest landscapes, local scale habitat patch configuration seems not to be very important for the Siberian flying squirrel but at a regional scale spatial spacing of larger landscape units such as old-growth forest areas or nature reserves is likely to have an effect on regional scale population dynamics. This suggests further that landscape configuration may be determined by long distance dispersers and, thus, by occasional but regular interchange of individuals between subpopulations.