Understanding Variability in Ecosystem Metabolism Measurements: the Important Role of Physics

Models of ecosystem metabolism of inland waters generally assume that diel changes in dissolved oxygen (DO) are driven by gross primary productivity (GPP) and respiration (R), after accounting for net diffusive fluxes. However, models usually do not explicitly account for most effects of physics. Using datasets from seventeen GLEON (Global Lake Ecological Observatory Network) lakes, we tested whether exogenous drivers that affect lake thermal stability and metabolic signal strength can explain variation and uncertainty in metabolism estimates. We used LN (lake number, a dimensionless index describing the balance between the stabilizing force of thermal stratification and the destabilizing force supplied by wind) and PAR (photosynthetically active radiation, 400-700nm) as metrics of physical stability and noise. LN serves as a proxy for “sensor footprint”, a concept we introduce to represent the integration volume represented by sensor measurements; PAR is a regulator of GPP and may alter both the sensor footprint and the metabolic signal to noise ratio. LN and PAR were significant predictors of uncertainty in GPP and R estimates; larger and more productive lakes were more sensitive to their effects. We identified and separated days when DO signals were potentially physically dominated and show that physically dominated days consistently resulted in higher uncertainty in GPP but lower uncertainty in R. However, physically dominated conditions affected parameter estimates in only about one quarter of systems. This work highlights the complexity of coupled physical-biological signals, and may be used to improve future studies of ecosystem metabolism and carbon cycling in inland waters.