Multi-scale variation in fine-root biomass in a tropical rain forest: A seven-year study

Understanding the dependency of ecosystem processes on spatial and temporal scales is crucial in current efforts to model ecosystem responses to global change. Here we present a case of nonlinear interactions between temporal and spatial scales in a high spatial- and temporal-resolution study of fine-root biomass responses to edaphic and climatic variation in a lowland tropical rain forest (La Selva, Costa Rica). The spatiotemporal variation in fine roots in this forest was considerably greater than that for aboveground live biomass and litterfall. Standing stocks of both live and dead fine roots varied strongly during a seven-year period (up to 10-fold) and across two edaphic gradients with different soil nutrient and water variation (up to four-fold). Fine-root biomass decreased with soil fertility and volumetric soil water content, but responses across a landscape gradient (decreasing with total soil P and K and increasing with N:P ratio between two Oxisols with different weathering) differed from those across a topographic gradient in older Oxisols (increasing with total Fe and Al and decreasing with Ca, Mg, and C:N ratio down the slopes). The spatial contrasts across the landscape gradient (but not in the topographic gradient) changed substantially through time, and they, in fact, disappeared entirely by the middle of the study interval. Shortterm monitoring of belowground standing biomass could thus lead to important biases when quantifying root responses. The positive time X gradient interaction in fine-root biomass across soil types (but not downslope) also indicates nonlinear changes in spatial patterns across temporal scales, so studies on temporal responses also need to be spatially explicit at narrow scales. This interaction also appears to be more complex than previously recognized: semester-averaged fine-root biomass was negatively correlated with soil water content in the preceding semester, but the relationship was restricted to residual Oxisols. To increase the accuracy of global carbon cycle models in the future, a critical complement to observations of ecosystem processes at regional and global scales will be quantifying these processes through multiple years and across the substantial edaphic gradients that typically exist within landscapes.