Models of litterfall production for Scots pine (Pinus sylvestris L.) in Finland using stand, site and climate factors

The aim of this study was to identify the most significant site, stand and climate factors affecting needle (LFneedle) and total (LFtotal) above-ground litterfall production and to develop multiple linear regression (MLR) models that can be used to reliably predict litterfall production in the boreal zone using readily available variables. Unlike most other litterfall production studies, we use climate data for the actual sites and annual litterfall values. A data set including 34 Scots pine stands located throughout Finland was compiled. The data for some stands covered a period of more than 30 years. The age of the stands ranged from 35 to >200 years and all were growing on upland, mineral soil sites. Stand mean annual LFneedle ranged from 22 g m−2 (northern Finland) to 157 g m−2 (southern Finland); corresponding values for LFtotal were 32 and 230 g m−2. Annual LFneedle production accounted between 49 and 75% of stand LFtotal production and explained 88% of the variation in LFtotal over all stands. There was considerable annual variation in litterfall production also within the same stand. The coefficient of variation in LFneedle in each stand ranged from 4 to 58% (mean = 19%) and from 3 to 39% (mean = 22%) for LFtotal. Both LFneedle and LFtotal were highly significantly (p < 0.01) and strongly correlated (Spearman) with latitude, stand basal area, effective temperature sum (ETS) of the current year and even higher with that of the previous year, and the previous years' July temperature. LFneedle had a weak negative, although significant (p < 0.05) correlation with stand age, but age was not significant for LFtotal. MLR models using latitude and stand basal area (also dominant tree height in the case of LFneedle) as predictive variables accounted for 82% of the variance in both LFneedle and LFtotal. The standard error of the estimate (SEest) was 12.6 g m−2 for LFneedle and 23.3 g m−2 for LFtotal. Latitude effectively described the climate at each stand but ignored the considerable within-stand variation in annual litterfall production. Using the annual values for the climate variables instead of latitude, 70% or more of the variation in both LFneedle and LFtotal in MLR models could be explained. The models are useful tools for predicting annual litterfall in mature Scots pine stands for use in soil organic and carbon models.