NEW ESTIMATES OF LIVE BIOMASS AND NET PRIMARY PRODUCTION OF RUSSIAN FORESTS: A FOOTPRINT OF CLIMATE CHANGE? A. Shvidenko 1 , D. Shepashenko 2 , S. Nilsson 1 , and A. Lapenis 3 1 International Institute for Applied Systems Analysis, A-2361 Laxenburg, Austria; shvidenk@iiasa.ac.at 2 Moscow State Forest University, Mytischi, Moscow Region; dschep@mimgroup.com 3 University at Albany, 218AS, 1400 Washington Ave., Albany NY, 12222, USA; andreil@albany.edu ABSTRACT The paper presents new estimates of live biomass (phytomass) and net primary production (NPP) of Russian forests for 1993 and 2003. These indicators are estimated based on forest inventory data and a specially developed semi- empirical modeling system. The latter contains regional models of growth by major forest forming species, multi- dimensional models of phytomass and models of biological production. It is shown that the fractional structure of forest phytomass substantially differs from previous estimates that indicated significant temporal trends of the share of aboveground wood (AGW), green part (GP) and belowground (BG) phytomass. The total forest NPP (of 307 g C m -2 yr -1 for 2003) is substantially higher than previously reported. These changes may be attributed to climatic change which was dramatic over the last four decades, particularly in Asian Russia. INTRODUCTION Live biomass (LB) and NPP are two crucial components of the terrestrial biota full carbon account. Available estimates of LB for Russian forests are obsolete. Data for NPP derived from vegetation models relate to potential vegetation and are biased. Inventory-based estimates of NPP have never been reported. The major objective of this study was to provide a new inventory of these indicators for all Russian forests for 1993 (763.5 x 10 6 ha) and 2003 (776.1 x 10 6 ha) aiming at the most accurate assessment that is currently possible, in order to quantify the recent dynamics of the structural component of LB and NPP and try to understand the supposed links of climate change to these dynamics. METHOD AND MATERIAL We used forest inventory data in the form of the State Forest Account for 1993 and 2003. These data contain areas and growing stock by dominant species, age, site indexes and relative stocking by ~2000 forest enterprises across the country. In order to convert these data into LB, a system of multi-dimensional nonlinear regression equations has been developed. The equations were calculated based on a database (DB) collected for Northern Eurasia forests (of ~3600 sample plots) that contain field measurements of LB by seven components: stem, branches (both over bark), bark, foliage, roots, understory (US) and green forest floor (GFF) by dominant species and ecoregion. The statistical accuracy of the regressions allowed estimating the forest LB at the national level with uncertainties <5% (confidential interval is 0.9). In order to estimate NPP, we developed a modeling system for assessing the biological productivity of forest ecosystems based on (1) models of growth and yield (~100 regionally distributed models were used), (2) the above models of LB, and (3) additional set of ecological parameters such as turnover of fine roots, life span of green parts, etc. [Shvidenko et al., 2005]. Age dynamics of LB and NPP by fractions are the outputs of this system. The statistically significant temporal trends in the dynamics of allometric ratios of phytomass components have been reported for Russian forests recently as a footprint of their acclimation to climate change [Lapenis et al., 2005]. Using our DB of LB measurements (which were done between mid-1950s and 2003) we calculated the temporal trend of the share of AGW, GP and BG LB to growing stock as a function of the age of stands and the time of measurements. Such models have been developed on a regional basis with four large regions and three groups of species (light coniferous, dark coniferous and deciduous), and in the aggregated form for the entire DB (see Fig. 1, data are normalized by 1983 values).