Tree-ring widths are good proxies of annual variation in forest productivity in temperate forests.

Tree rings have long been used to calibrate the net primary production (NPP) time-series predicted by process-based models, based on an implicit assumption that ring-width indices (RWI) can well reflect temporal NPP change. However, this assumption has seldom been tested systematically. In this study, 36 plots were set in three forest types from four sites along a latitudinal gradient in northeast China. For each plot, we constructed chronologies and stand NPP of the past 20 years to examine: is RWI a good proxy of inter-annual variation of forest NPP for different forest types under different climate? If it is, why? Our results indicate that RWI was closely related to stand NPP in most cases, and could be used as a good proxy of NPP in temperate forests. Standard and arstan chronologies were better related to NPP series than residual chronology. Stand NPP time-series were mainly determined by large trees, and the correlation between RWI and NPP was also higher for larger trees. We suggest that large trees and dominant species of canopy layer should be sampled for chronology construction. Large trees are major contributors of forest biomass and productivity, and should have priority in forest conservation in a rapid-warming world.