DIRECTIONS OF LANDSCAPE CHANGE (1741-1993) IN VIRESTAD, SWEDEN - CHARACTERIZED BY MULTIVARIATE-ANALYSIS

The objective of the paper is to demonstrate the value of the multivar iate method, Principal Components Analysis (PCA) on data relating to l andscape elements, to determine the relative directions of change and illustrate landscape dynamics. Data were retrieved from retrospective studies based on interpretation of aerial photographs and old cadastra l maps at three time periods, 1741-1811, 1946 and 1993 covering 56 sel ected hamlets in a parish in Smaland, south Sweden. To carry out the a nalysis, data were summarised into five principal land cover groups; ( I) arable and grassland composite; (2) open semi-natural grasslands; ( 3) wooded grasslands; (4) water; and (5) forest/remaining land. During the 200-year period investigated, the landscape shifted in most hamle ts from a more open rural landscape with fields and areas of semi-natu ral grassland vegetation, to a non-rural landscape with closed forests and fragmented rural areas dominated by cultivated grasslands. The ra nge and distribution of hamlets along the principal axes over time ind icate a reduction of complexity, and suggest a negative relationship b etween the degree of open agricultural landscape and the relative prop ortion of semi-natural and wooded grasslands. PCA was shown to be a us eful tool to simplify the complex landscape data sets and indicate the character of change in terms of dynamics of spatial organisation, and the direction of change in the rural landscape of Virestad. In the an alysis, three hamlet types with diverging development were distinguish ed; (A) major change from a semi-open rural landscape to a closed non- rural forest landscape; (B) intermediate change, where hamlets were al ready closely related to the closed forest type landscape from the beg inning; and (C) intermediate change, where hamlets have remained close r to a relatively open rural landscape over the 200 years of the study period. (C) 1997 Elsevier Science B.V.