Fine scale heterogeneity of soil resources in a wet tropical forest

Background/Question/Methods

Spatial heterogeneity of above- and below-ground resources could contribute to tree species distributions, coexistence in communities, and population persistence.  Yet estimates of resource availability and heterogeneity are strongly dependent upon the spatial scale at which they are measured.  While both irradiance and soil resources may be extremely spatially heterogeneous across landscapes, only variability in irradiance has been measured at a fine seedling/sapling-level scale.  Soil resources, however, have not been measured at a similarly fine spatial scale that is commensurate with the soil environment experienced by tree seedlings.  The objectives of this research were to characterize spatial autocorrelation of soil resources in a wet tropical forest, compare resource heterogeneity within versus between sites, and determine the appropriate spatial scale to measure soil resources to understand tree seedling responses.  Five study sites were sampled in a tropical wet forest at La Selva Biological Station, Costa Rica. To represent landscape-level occurrence of soils, three study sites were sampled on residual soils and one site each on older and recent alluvial soils in old growth forests; sites were established to span a soil fertility gradient.  We measured a broad array of soil nutrients including nitrate, ammonium, phosphate, total phosphorus, aluminum, potassium, magnesium, and calcium. 

Results/Conclusions

Availability of aluminum, potassium, magnesium, and calcium soil concentrations exhibited strong spatial autocorrelation. To compare fine seedling-level scale versus plot level, sites were initially analyzed every 2m versus every 10m.  When intensively analyzed (every 2m), the average range of spatial correlation for both nitrate and ammonium was relatively short, below 15-20 m. This might help explain the perceived lack of spatial association seen in those analyses using a coarser grid (plot level, every 10m).  Linear interpolations of soil resources at every 2 m along a 200 m belt transect were also poor predictors of actual soil resource levels.  For all soil resources, except phosphate, variation within a site was approximately equivalent to variation across sites, in spite of the fact that the sites were on soils of different origins (alluvial versus volcanic) and ages; phosphate availability, however, differed significantly among sites.  These results suggest that to understand seedling responses to soil resources – and by the extension, the potential for species to partition gradients of soil resource availability --  below-ground resource availability should be measured at a fine, seedling-level spatial scale.