Of the 100+ landslides that occurred in the LEF as a result of the April 17-19, 2003 storm event, approximately 69 were road-related (within 200 m of a road; Larsen 1995), and were ≥12 m2 in area and <55o slope (A. Shiels, unpublished data). We randomly chose 30 of those 69 landslides for our study. The landslides ranged in elevation from 152 m to 825 m a.s.l. (see Appendix 1 in Shiels et al. 2008), spanned tabonuco, palo colorado, palm, and cloud forest types, and were found on areas underlain by both volcaniclastic (n = 8) and diorite (n = 22) soil types. In October 2003, we randomly located a 1 x 1m plots within the landslide chute (the central area of highest erosion within a landslide; Guariguata 1990) of each of the 30 landslides. The chute portion of the landslide was selected because it was present on all chosen landslides (often road-related landslides have the lower deposition zone removed in order to clear the roadways), and was not too steep to access (the upper, slip-face zones often have near-vertical slopes). The plots were at least 1 m (mean ± SE = 2.54 ± 0.21 m) from any forest-landslide edge. Within each plot on each of the 30 landslides, we measured all of the parameters for this study, which included: slope stability and litterfall from December 2003 to May 2004 (8-13 months following landslide formation); and in May 2004, we measured microtopography, soil characteristics (organic matter, particle size, nitrogen, and water holding capacity), elevation, distance to forest edge, canopy cover (measured with a densiometer), and vegetation colonization.
Sediment runoff was measured at the base of each plot using a 100 cm x 8 cm x 7 cm deep, open-top plastic rain gutter placed perpendicular to the slope and tilted to allow flow into a plastic collection bucket. We removed and obtained wet mass of all sediment from the gutters and buckets every two weeks from December 2003 to May 2004 (24 weeks), then determined total sediment dry mass by taking subsamples of the sediment wet mass and calculating wet/dry mass ratios.
Because microtopography of the soil surface can affect plant colonization, we calculated a roughness index (Saleh 1993) of each litter removal plot during May 2004. We measured the depth to the soil surface from an imaginary horizontal plane placed on the highest emergent point from the plot surface and parallel to the plot slope. Measurements were made every 10 cm along three horizontal transects across each plot (25, 50 and 75 cm from the bottom edge; total = 33 measurements per plot).
Organic matter measurements included litterfall (litter removal plots only) and SOM (both control and litter removal plots). Litterfall was collected from each plot, rinsed to remove sediment and dried at 45o C before weighing. This process was repeated every two weeks from October 2003 to May 2004. Soil organic matter (SOM; % loss on ignition), was determined from three soil cores (1.9 cm diameter and 10 cm deep) taken in May 2004 from random locations within each plot. Soils from these cores were also analyzed for particle size (% sand and clay; Sheldrick and Wang 1993) and total nitrogen (Kjeldahl digestion followed by colorimetric analysis; Alpkem 1992). An additional soil core (7 cm diameter to 10 cm depth) was taken from each plot in May 2004 in order to determine the water-holding capacity of the soil by saturating each soil sample, allowing it to drain for 10 min before weighing, and finally drying (to a constant mass at 105o C) before reweighing each sample.
Plant colonization and early successional development on the 30 landslides were evaluated by measuring vegetation cover and biomass from each plot in May 2004. Because cover estimates are vertical projections onto the ground surface, values can exceed 100%. All aboveground plant biomass (live or standing dead) was removed from each plot (litter removal and control) and dry mass was determined for the same six plant categories used for vegetation cover. Only plants rooted in the plot were harvested. Belowground biomass (coarse roots only) was also estimated in May 2004 by sampling then pooling two cores in each plot (7 cm diameter, 16 cm depth, 616 cm3) and manually separating all roots and fern rhizomes (live or dead) >1 mm diameter before drying them to a constant mass.