Abstract Aim The Holocene spread of Picea abies in Fennoscandia is well established from many sites and thus provides an opportunity for detailed study of the dynamics of tree spread and population expansion. Early- and mid-Holocene macrofossil evidence for presence
of P. abies in Fennoscandia has questioned traditional interpretations of the timing and direction of its spread. This paper aims to determine when, from where and by which pathways P. abies spread into Fennoscandia. Understanding the character and dynamics of this spread may
give insight into the general understanding of Holocene tree spread. Location The north-western distribution of P. abies in Europe, including Norway, Sweden, Finland, Estonia, Latvia, Lithuania, north-western Russia, parts of Byelorussia and Poland. Methods
Holocene pollen diagrams with independent dating control were collected from this region. The timing of the onset of the continuous curve, the timing of the rise of the curve, the first appearance of frequencies of 1%, 3%, 5%, and 10%, as well as timing and the maximum amount of P.
abies pollen, was obtained from these pollen diagrams. A GIS analysis was used to display the data and interpolate ages over the area under investigation. Results Maps are presented showing a clear ESE to WNW trend in the spread of P. abies for all characters interpolated.
The timing of the rise of the curve was difficult to use as sites east of the Baltic have slowly rising P. abies frequencies while the western sites often show a rapid rise. Main conclusions The spread of P. abies in Fennoscandia and adjacent areas can be separated
into two phases: (i) A rapid early Holocene spread out of Byelorussia and northern Russia at low population density giving rise to small outpost populations, possibly as far west as the Scandes Mountains. (ii) A mid- to late Holocene front-like spread at high population densities moving from
east to west into the Baltic Republics and Finland, into northern Scandinavia and then moving south and west towards its present-day distributional limits.