It was examined whether the present site classification method, and especially its applicability to site productivity estimation, could be improved in upland Scots pine (Pinus sylvestris L.) forests in Southern Finland by developing a classification key based on Two-way Indicator Species Analysis (TWINSPAN), and/or by inclusion of soil texture, stoniness and the humus layer depth more closely in the classification method. TWINSPAN clusters (TW) explained 71%, and forest site types (FST) 64% of the variation in site index (SI) (H100). When soil texture (TEXT) was added to the regression model, the explanatory power increased to 82% (SI = TW + TW * TEXT) and to 80% (SI = FST + FST * TEXT), respectively. Soil texture alone explained 69% of the variation in site index. The influence of stoniness on site index was significant (P <0.05) on sorted medium sand soils and on medium and fine sand moraine soils. The thickness of the humus layer (2–6 cm) was not significantly (P=0.1) related to site index.
It is suggested that the proposed TWINSPAN classification cannot replace the present forest site type system in Scots pine stands in Southern Finland. However, the TWINSPAN key may be used to aid the identification of forest types. The observation of dominant soil texture within each forest type is recommended.
The PDF includes an abstract in Finnish.