Plant species. We used five common plants from the mature tabonuco forest as well as the mid-successional forest. Species selection was based primarily on nitrogen and phosphorus content, leaf structure, presence or absence of latex, and plant phylogenetic relationships. The five leaf species used in these experiments were: Inga vera and I. fagifolia in the Leguminosae; Croton poecilanthus and Sapium laurocerasus in the Euphobiaceae; and Manilkara bidentata in the Sapotaceae. The plant species were paired based on their physical and chemical characteristics (Tables 1, 2) to facilitate comparisons. Species in the first pair, I. vera and I. fagifolia, share the same family and genus and have high N and lignin concentrations, but they differ slightly in their levels of polyphenols. Species in the second pair, M. bidentata and C. poecilanthus, share similar nutrient concentrations, leaf structure and latex, but they belong to different families. Species in the third pair, C. poecilanthus and S. laurocerasus, share the same family and have latex, but they differ in nutrient and lignin concentrations as well as leaf structure. Sapium laurocerasus leaves are thin and smooth while C. poecilanthus leaves are thicker and have a heavier cuticle with peltate hairs. Thickness of the leaves, epidermal cells, epidermal cell walls, and waxy cuticles were measured on hand-cut cross-sections using an ocular micrometer at 200-600x (five leaves per species). Ten fresh leaves per species were weighed and their areas were estimated using a grid. Water uptake (g water/cm2) was then determined by submerging the same leaves in deionized water for two hours and re-weighing (Taylor & Parkinson, 1988b).Leaf litter collection. Only recently fallen senescent leaves of the selected species were collected. These leaves represent the normal substrates for primary fungal decomposers. Most leaves were collected the same day they abscised, but some senescent leaves where collected directly from their trees. Leaf collection took place during April and in early May for the first experiment, and at the end of July for the second set of experiments. Leaves of each species were collected from at least five trees. For field experiments, leaves were dried at 50ºC for 72 h. The 50°C temperature was used to avoid major changes in plant chemistry and to control for differences in initial water content between leaf species (W. Silver, personal communication). Additional leaves were collected for chemical analyses and were dried according to each laboratory's protocols as described below. Effects of leaf structure and quality on decomposition. Litter bags (20 x 20 cm) made with 2 mm mesh plastic screening were used for these experiments. Litter bags containing three grams of oven-dried (50°C) leaves of a single species were placed in a common plot. The first experimental sets were decomposed from May through July 2000. A total of 45 bags were used, nine bags per plant species.Bags were placed in the field under a tree of their own species for the first week to allow colonization by their normal decomposers, and subsequently moved to the common plot located in mature tabonuco forest. Three replicates per plant species were collected after three, six and twelve weeks of decomposition. After each collection, bags with the decomposed leaves were oven dried at 65°C for 72 h and weighed to calculate rates of decomposition (difference between the initial and the final dry weight). We used correlation analyses to determine the relationship between leaf quality and decomposition and correlations among leaf quality factors. Another set of bags, five replicates for each species, was set in the common plot from the end of August to November 2000. This set was collected after 12 wks. and was used to compare the effects of dry versus wet periods on decomposition. This set was also used to compare decomposition of leaves in the common plot with leaves simultaneously placed under their source trees to determine the effect of microsite on decomposition. Effect of site on decomposition. We used two types of sites: the common plot in tabonuco forest as described above was used as a foreign site, and the forest floor under trees of the same species as those in the litterbags as a ˜source site". The experiment was conducted only during the wet period. In the ˜source site" set, each plant species was represented by five trees located at least 100 apart. Under each of these five trees per species, we placed three litterbags of the same species; one bag was collected on each sample date. A total of 25 trees and 75 leaf litterbags were used in the source tree decomposition experiment. A total of 25 bags were placed in the common plot, five of each leaf species. Five replicate bags of each species were retrieved after 3, 6 and 12 wks of decomposition from under the source trees, and after 12 weeks from the common plot. After collection, litter was processed as described above. In addition, we compared early stage leaching losses in the two types of sites. The amount of rainfall during the first phase of decomposition (3 wks) did not differ significantly between the dry and wet period experiments described below (207 and 197 mm in the dry and wet period experiments, respectively). We were therefore able to compare initial mass loss attributed primarily by leaching between sites (two-way ANOVA), even though these experiments occurred at different times. Influence of wet and dry periods on decomposition rates. For these experiments, comparisons were made of decomposition rates during the dry and wet periods in the common plot. The dry period was from May through July 2000 and the wet period from August-November 2000. Five replicates were used in the experiment during the wet period, and three replicates were used per collection time during the dry period.
