To measure major components of net primary production (NPP) and their annual and spatial variation, we have established stratified replicated plots across the major edaphic gradients of La Selva's old-growth forest. In these plots we obtain annual rates of stand-level tree biomass increment and production of fine litterfall, coarse-woody debris, and fine root biomass. These data allow us to examine temporal and spatial variation in major components of NPP.

The 18 plots were sited using the soils (Sancho & Mata 1987) and topography (OTS unpublished data) coverages of the La Selva GIS, as follows (see map):

a) 6 plots on (flat) inceptisols (Experimental and Holdridge
Consociations; Sollins et al. 1994)
bi) 6 plots on plateau sites on ultisols (Matabuey and
Jaguar Consociations; Sollins et al. 1994)
ci) 6 plots on slope sites on ultisols: 3 on slopes of NW
aspect, 3 on slopes of SE aspect.

Plot size was based on the results of preliminary sampling to estimate the relation between plot area and variance in total basal area. Final plot sites is 0.5 ha, which is considerably larger than the mean canopy gap size at La Selva (161 m2; Sanford et al. 1986). Each study plot includes a mixture of forest phases (gap to mature) andmore than 200 stems >10cm in diameter. Stems this size account for > 90% of forest basal area (Lieberman & Lieberman 1994). Within the edaphic sampling regions, plots were sited using randomly-drawn coordinates from the La Selva grid systemminimizing biases due to non-representative stand selection in terms of forest phase or tree size distribution (c.f. Phillips & Gentry 1994; Clark et al. 1995).

All stems > 10 cm diameter (dbh or above buttresses) were tagged and identified to species in 1997. We measure diameter of all stems > 10 cm (protocols of Clark & Clark 1992; mean measurement error + 0.3 mm) annual during the months of September - November. Allometric (diameter/biomass) equations for neotropical wet forest trees (Brown et al. 1989, Brown & Iverson 1992, Lieberman & Lieberman 1994) are used to estimate annual increments in aboveground tree biomass in the plots.

Soil moisture is monitored in all plots biweekly. Initial soil sampling and analysis (pH, exchangeable bases, total and extractable P) were performed to characterize each plot.

In TRF, fine litterfall is a significant component of aboveground annual net production, typically equaling or substantially exceeding wood production (Parker 1994). We measure fine litter production using 0.5m2 traps (5/plot, bulked for each plot at each collection), sampled initially at biweekly intervals. Samples are separated into leaves, twigs, and reproductive structures, and these components are dried and weighed (with a sample from each plot stored for possible nutrient analyses and species separations).

Stand-level LAI will be estimated biannually (wet season/dry season) in all plots by taking canopy photographs with a fish-eye lens at permanent sample points. The images are analysed with the software program WINPHOT (ter Steege 1996).

At plot initiation we assessed total coarse woody debris (CWD) present on the plots by mapping fallen branches and logs >10 cm diameter. Annual CWD input are assessed at each annual census by mapping and sub-sampling for biomass (cf. Sollins 1982, Harmon et al. 1987, Bingham and Sawyer 1988, Spies et al. 1988).