LANDSCAPE SCALE EVALUATION OF UNDERSTORY LIGHT AND CANOPY STRUCTURE -METHODS AND APPLICATION IN A NEOTROPICAL LOWLAND RAIN-FOREST

Light is a key resource controlling tree regeneration in the understor y of closed-canopy old-growth forests. To evaluate the distribution of understory light environments at a landscape scale, we used stratifie d random sampling in a 500-ha stand of Costa Rican tropical rain fores t. Fifteen 100 m long transects were placed using random coordinates w ithin two soil-geomorphology units (flat alluvial terraces and dissect ed ridge-slope-swale terrain). At 2.5-m intervals we measured canopy h eight and slope angle, classified topographic position, and took canop y photographs with a fish-eye lens at 1 and 3 m above the ground (and at 0.6 and 5 m height at five stations per transect). Photographs were analyzed for global site factor (GSF), which is analogous to the perc entage of full sun radiation reaching a point. Canopy height and GSF a t 1 and 3 m above the ground were significantly autocorrelated (Moran' s I) at 2.5-m intervals. The autocorrelation rapidly declined at great er intervals, reaching nonsignificance at ca. 20 m. Both canopy height and GSF at 3 m height had a weak tendency for negative autocorrelatio ns at intervals of 25-50 m. Median canopy height (615 stations) was 23 m (range 0-37). Caps (canopy height less than or equal to 2 m, Brokaw 1982) were only 1.5% of sample points. Caps were more frequent on ste ep slopes than on terraces, ridgetops, swales, and gentle slopes. Cano py height varied significantly across this topographic gradient. At al l four heights (0.6, 1, 3, and 5 m) median GSF was less than or equal to 2.4%. GSF values > 8% accounted for only 3% of the total sample (N = 1380). GSF was only weakly negatively correlated with canopy height and the relation was not monotonic. Under canopies 13-19 m tall, nearl y all GSF values were less than or equal to 5%. Higher GSFs were more frequent under both shorter and taller canopies. Given the observed va riance in GSF and canopy height, 100-200 points separated by intervals of greater than or equal to 20 m are necessary to measure the forest- wide means of these variables to +/- 10%. We discuss implications of t hese results for current approaches to modeling understory light based on canopy characteristics We compared the random background of light environments from 1-3 m above the ground at La Selva with those occupi ed by saplings of pioneer and nonpioneer tree species. The two pioneer s (Cecropia spp.) occurred in microsites significantly brighter than r andom sites, while sapling microsites of all five nonpioneer species w ere significantly darker than random. Comparing the landscape-scale di stribution of key resources with species' actual distributions at simi lar scales offers a quantitative method for assessing plant life histo ries within and among forests.