1 Populations of Rubus saxatilis were investigated between 1988 and 1991 in a study area (c. 2.5 km2) in central Sweden. Different phases in the life cycle were studied: flowering, pollination, fruit-set, fruit-removal and seedling recruitment. Identification of genets was made by use of RAPD (random amplified polymorphic DNA). Information concerning these life-cycle phases were then used to infer processes in the population dynamics of R. saxatilis. 2 A path analysis suggested a strong effect of fruit-set on total patch fruit production. Fruit-set varied between 6.6% and 8.3% (yearly averages) and was influenced by distance to nearest flowering conspecific patch. Isolated patches tended to have a low fruit-set. Size of patches was of minor importance for fruit-set and fruit-production, whereas intrapatch ramet density influenced fruit production positively via an effect on the proportion of ramets that flowered. 3 A pollination experiment confirmed that R. saxatilis is self-incompatible. Within-patch pollinations indicated that isolated patches comprise single genets (or incompatibility types). Between-patch pollination indicated that fruit-set may be pollen limited, at least some years. Deficiency of compatible pollen (or 'partners') is the most likely mechanism behind the distance-effect on fruit production found in the path analysis. 4 A bagging experiment suggested that c. 50% of the fruits (drupelets) were removed by animals. No relationship was found between fruit removal rate and either fruit production of patches or distance to neighbour patch. 5 Seedlings of R. saxatilis were carefully searched for during four years in three different localities 'coniferous forest', 'deciduous forest' (DF) and 'mixed semi-open woodland' (MSOF). Seedlings were found only during one year and in one locality (MSOF). Ramets from carefully mapped 20-m x 30-m plots in localities DF and MSOF (one plot per site) were sampled for genet identification with RAPD analysis. In DF, 20 ramets were found to belong to 10 different genets. In MSOF, 24 ramets belonged to eight different genets. In addition, a sample of three ramets from an isolated road-side patch indicated that it consisted of one genet. 6 The dynamics of the genet population of R. saxatilis in the study area was interpreted by a source-sink concept. Source populations represent local clusters of genets with high fruit production which enhances further local recruitment in source populations and provide the surrounding sink areas with propagules. Sink populations represent low-density populations comprising isolated genets suffering from partner-limited fruit production. These isolated genets may be loci for future colonization. The source-sink structure implies an apparent positive genet density effect on recruitment, but to what extent this effect is due to variation in the local 'quality' of sites or is a real density effect, remains unresolved.
