Long-term rainfall and discharge data from the Luquillo Experimental Forest (LEF) were analysed to develop relationships between rainfall, stream-runoff, and elevation. These relationships were then used with a Geographic Information System (GIS) to determine spatially-averaged, mean annual hydrologic budgets for watersheds and forest types within the study area. Model estimates indicate that a total of 3864 mm/yy (444 hm3) of rainfall falls on the forest in an average year. The Tabonuco, Colorado, Palm and Dwarf Forest types receive an estimated annual rainfall of 3537, 4191, 4167, and 4849 mm/yy, respectively. Of the average annual rainfall input, 65% (2526 mm/yr) is converted to runoff and the remainding 35% (1338 mm.yr) is lost from the system by evapotranspiration and other abstractions. In comparison to other tropical forests, the LEF as a whole has more evapotranspiration than many tropical montane forests but less evapotranspiration than many lowland tropical forests.

Dataset Methods:

Field Collection:

Rainfall and throughfall collection
The rainfall and throughfall measured in this study were collected and measured in the same manner for the duration of the study, and in accordance with our previous publications (Scatena, 1990; Schellekens et al., 1999; Holwerda, 2006). Bulk rainfall and throughfall were collected weekly (i.e. every Tuesday morning) and occasionally before and after major storms. Total rainfall was collected in a 25 m above canopy walk-up tower that is located on the divide between the two catchments and at an elevation of 361 m above sea level. Throughfall was measured throughout the watershed using 20 to 35 randomly placed but fixed gauges that were placed across the ridges, hill-slopes, gaps, and stream channels of the watershed (Scatena, 1990). The rainfall collector and each throughfall collector had identical 143 cm2 funnels. As many as 35 collectors were operated at any given time for the time series presented here. During Hurricane Hugo in 1989, the canopy tower that held the climate station and rainfall collector was toppled, but most of the throughfall collectors remained intact. Those that were destroyed were randomly re-located within 10 m of their original site and throughfall was collected without interruptions. The meteorological tower was also replaced after a few months.
Lab Analysis:

Chemistry
During every collection, bulk rainfall and throughfall were collected for chemical analysis. Water for the rainfall analysis was collected from the above-canopy rainfall collector. The throughfall sample was a composite of water collected in eight collectors. These eight throughfall collectors were selected at the beginning of the study because their mean throughfall volume and conductivity was similar to the mean of all the bottles and therefore considered representative of the site. These collection bottles were cleaned or replaced on a weekly basis and contained filters to prevent frogs and litter from entering the bottles.
Water samples were delivered to the laboratory on the same day they were collected. Chemical analysis was conducted in the same manner as previous studies of the LEF (McDowell et al., 1990; McDowell and Asbury, 1994; McDowell, 1998). Protocols and the original data are available on the Luquillo LTER web-page:

(http://luq.lternet.edu/data/lterdb20/metadata/lterdb20.htm)

In the laboratory, pH and conductivity were measured following the procedures specified by NADP (1984) and McDowell et al., (1990). Samples were filtered using pre-combusted glass fiber filters (Whatman GF/F). Until 1997, samples were held refrigerated for analysis, with a sub-sample for ammonium analysis preserved by acidification with sulfuric acid (McDowell et al., 1990). After 1997, samples were stored frozen until analysis for all constituents except silica, which was analyzed on a refrigerated subsample. During the first nine years of the study, most samples were analyzed at the University of Puerto Rico. After 1997, all samples were analyzed at the University of New Hampshire. Silica (phospho-molybdate), phosphorus (ammonium molybdate), and ammonium (phenol-hypochlorite) were analyzed throughout the study period using spectrophotometric methods using a Technicon AA II or Lachat Quickchem. Cations were analyzed with atomic absorption spectrophotometry from 1988-1994, and with ion chromatography from 1994 on. Anions were measured with ion chromatography. Dissolved organic carbon and nitrogen were measured using persulfate digestion (McDowell et al., 1987; Solorzano and Sharp, 1980) prior to 1997, and with high temperature Pt-catalyzed combustion after 1997 (Merriam et al., 1996). Cross-lab comparisons and analysis of samples using the different techniques indicated that comparable results were obtained with different laboratories and methods (e.g. McDowell et al., 1990; Merriam et al., 1996).