<p>There has been a long-term decline in number of cyclic vole populations in boreal Sweden since the 1970s. Several hypotheses have been suggested to explain this decline. Commonly for <i>C. glareolus</i>, <i>C. rufocanus</i> and <i>M. agrestis</i>, the decline has followed upon an increased frequency and severeness of winter declines and has shown up as a drop in spring densities. The spring decline is most pronounced for <i>C. rufocanus</i>. In contrast to other voles, <i>C. rufocanus</i> also show a decline in fall densities, suggesting some additional disturbance in this species. Habitat fragmentation has been suggested as such an additional disturbance and in this thesis the effect of habitat fragmentation on <i>C. rufocanus</i> is explored.</p><p>At first the sampling method was evaluated i.e. whether the decline could be due to destructive sampling when the method in use in the long-term monitoring is snap-trapping. This resulted in a rejection of the destructive sampling hypothesis as a possible cause behind the decline in <i>C. rufocanus</i>. Habitat preference revealed that three habitats at the local scale (trap station) were high quality habitats for <i>C. rufocanus</i>: forest of moist and wet/hydric dwarf-shrub type, in addition to forest/swamp complexes rich in dwarf-shrubs. The occurrence of <i>C. rufocanus</i> at the landscape scale was positively correlated with the amount of boulder fields and a low degree of fragmentation of old-growth pine forests. There was considerable local variation in the decline in vole density among the 58 1-ha sampling plots, with respect to both density and timing of the decline, which suggested that habitat destruction outside sampling plots might be involved. Overall, clear-cuts had a negative influence on vole densities at both the local and landscape scale. A multiple regression analysis suggested that having both a high quality habitat at the local scale and a high proximity among xeric-mesic mires and a low connectivity among clear-cuts at the landscape scale were important for the occurrence of <i>C. rufocanus</i>.</p><p>Initial analysis at the landscape scale were based on landscape data collected from 2.5 x 2.5 km areas centred on the individual vole sampling plots. Further investigations, however, on the patch level suggest that focal forest patch size and quality was of major importance in determining occurrence and persistence of <i>C. rufocanus</i>. Although not tested formally in these studies, the habitat fragmentation hypothesis has so far received support. Currently <i>C. rufocanus</i> seems to be affected negatively by too low patch sizes of suitable habitats in the surrounding landscape suggesting that the amount of suitable habitats could already be below the fragmentation threshold. However, this has to be evaluated further. Work is in progress to establish time-series over local landscape changes, and to evaluate if such changes have been associated with local declines of <i>C. rufocanus</i> and whether habitat loss, true habitat fragmentation or both have been influential.</p>

There has been a long-term decline in number of cyclic vole populations in boreal Sweden since the 1970s. Several hypotheses have been suggested to explain this decline. Commonly for C. glareolus, C. rufocanus and M. agrestis, the decline has followed upon an increased frequency and severeness of winter declines and has shown up as a drop in spring densities. The spring decline is most pronounced for C. rufocanus. In contrast to other voles, C. rufocanus also show a decline in fall densities, suggesting some additional disturbance in this species. Habitat fragmentation has been suggested as such an additional disturbance and in this thesis the effect of habitat fragmentation on C. rufocanus is explored. At first the sampling method was evaluated i.e. whether the decline could be due to destructive sampling when the method in use in the long-term monitoring is snap-trapping. This resulted in a rejection of the destructive sampling hypothesis as a possible cause behind the decline in C. rufocanus. Habitat preference revealed that three habitats at the local scale (trap station) were high quality habitats for C. rufocanus: forest of moist and wet/hydric dwarf-shrub type, in addition to forest/swamp complexes rich in dwarf-shrubs. The occurrence of C. rufocanus at the landscape scale was positively correlated with the amount of boulder fields and a low degree of fragmentation of old-growth pine forests. There was considerable local variation in the decline in vole density among the 58 1-ha sampling plots, with respect to both density and timing of the decline, which suggested that habitat destruction outside sampling plots might be involved. Overall, clear-cuts had a negative influence on vole densities at both the local and landscape scale. A multiple regression analysis suggested that having both a high quality habitat at the local scale and a high proximity among xeric-mesic mires and a low connectivity among clear-cuts at the landscape scale were important for the occurrence of C. rufocanus. Initial analysis at the landscape scale were based on landscape data collected from 2.5 x 2.5 km areas centred on the individual vole sampling plots. Further investigations, however, on the patch level suggest that focal forest patch size and quality was of major importance in determining occurrence and persistence of C. rufocanus. Although not tested formally in these studies, the habitat fragmentation hypothesis has so far received support. Currently C. rufocanus seems to be affected negatively by too low patch sizes of suitable habitats in the surrounding landscape suggesting that the amount of suitable habitats could already be below the fragmentation threshold. However, this has to be evaluated further. Work is in progress to establish time-series over local landscape changes, and to evaluate if such changes have been associated with local declines of C. rufocanus and whether habitat loss, true habitat fragmentation or both have been influential.