The data were obtained at 23 forest permanent plots in 21 sites. These plots cover major climate zones and biogeographic regions in Japan (Ministry of the Environment 2001, http://www.env.go.jp/press/file_view.php?serial=2872&hou_id=2908) (Fig. 1, see also Ishihara et al. 2011). Furthermore, these plots cover major four forest types found in Japan: evergreen conifer forest (EC), broadleaf and conifer mixed forest (BC), deciduous broadleaf forest (DB), and evergreen broadleaf forest (EB). We classified each plot into one of the four forest types by the dominance of evergreen broadleaf, deciduous broadleaf and conifer tree (Ishihara et al. 2011). Most of these forests are old-growth or old secondary forests but some are secondary forests that are younger than 100 years. We classified forests into one of three age categories: old-growth (OG), old secondary (OS), and secondary (S) (Ishihara et al. 2011; see SiteList.csv, Appendix.pdf). We assigned to each plot an alphanumeric code as [abbreviated site name]-[forest type code][plot number] (see SiteList.csv). For example, AS-DB2 represents the second plot of deciduous broadleaf forest (DB) in the site, Ashoro (AS). Mean annual temperature, mean annual precipitation, and mean maximum snow depth during 1971 to 2000 were extracted from the database Mesh Climate Data 2000, distributed by the Japan Meteorological Agency (2002). The database provides climate variables estimated at the spatial resolution of 1-km. Mean annual temperature was corrected for altitudinal difference between the plot and the 1-km grid mean by the lapse rate of 0.6°C·100m-1. Snow depth from the database might underestimate or overestimate the snow depth of the plots because snow depth is highly spatially heterogeneous. Therefore, we provide snow depth reported in other publications or according to personal observations in Appendix.pdf.

One or two permanent plots were established in each site for litter collection. The sizes of the plots are usually 1 ha (100×100 m) ranging from 0.12 to 1.2 ha (see SiteList.csv). The shapes of the plots are normally squares or rectangles but some plots have different shapes because we avoided edges of the forest that face artificial forest, roads, or buildings. Each plot was usually divided into 10×10 m grid cells. Tree census has been conducted in each plot at 1-year or 5-years intervals (see Ishihara et al. 2011).