Prediction of stream nitrogen and phosphorus concentrations from high-frequency sensors using Random Forests Regression.

Stream nutrient concentrations exhibit marked temporal variation due to hydrology and other factors such as the seasonality of biological processes. Many water quality monitoring programs sample too infrequently (i.e., weekly or monthly) to fully characterize lotic nutrient conditions and to accurately estimate nutrient loadings. A popular solution to this problem is the surrogate-regression approach, a method by which nutrient concentrations are estimated from related parameters (e.g., conductivity or turbidity) that can easily be measured in situ at high frequency using sensors. However, stream water quality data often exhibit skewed distributions, nonlinear relationships, and multicollinearity, all of which can be problematic for linear-regression models. Here, we use a flexible and robust machine learning technique, Random Forests Regression (RFR), to estimate stream nitrogen (N) and phosphorus (P) concentrations from sensor data within a forested, mountainous drainage area in upstate New York. When compared to actual nutrient data from samples tested in the laboratory, this approach explained much of the variation in nitrate (89%), total N (85%), particulate P (76%), and total P (74%). The models were less accurate for total soluble P (47%) and soluble reactive P (32%), though concentrations of these latter parameters were in a relatively low range. Although soil moisture and fluorescent dissolved organic matter are not commonly used as surrogates in nutrient-regression models, they were important predictors in this study. We conclude that RFR shows great promise as a tool for modeling instantaneous stream nutrient concentrations from high-frequency sensor data, and encourage others to evaluate this approach for supplementing traditional (laboratory-determined) nutrient datasets.

Implications of irradiance exposure and non-photochemical quenching for multi-wavelength (bbe FluoroProbe) fluorometry.

Multi-wavelength fluorometers, such as the bbe FluoroProbe (FP), measure excitation spectra of chlorophyll a (Chl-a) fluorescence to infer the abundance and composition of phytoplankton communities as well as the concentration of chromophoric dissolved organic matter (CDOM). Experiments were conducted on laboratory cultures and on natural communities of freshwater phytoplankton to determine how the response of phytoplankton to high irradiance might affect fluorometric estimates of community composition and concentrations of Chl-a and CDOM. Cultures of a representative cyanobacterium, bacillariophyte, synurophyte, cryptophyte, and chlorophyte revealed changes in Chl-a excitation spectra as irradiance was increased to saturating levels and non-photochemical quenching (NPQ) increased. The degree of change and resulting classification error varied among taxa, being strong for the synurophyte and cryptophyte but minimal for the cyanobacterium. Acute-exposure experiments on phytoplankton communities of varying taxonomic composition from five lakes yielded variable results on apparent community composition. There was a consistent decrease in CDOM estimates, whereas Chl-a estimates were generally increased. Subsequent exposure to low PAR relaxed NPQ and tended to reverse the effects of high irradiance on composition, total Chl-a, and CDOM estimates. Relaxation experiments on near-surface communities in a sixth, large lake, Georgian Bay, showed that total Chl-a estimates increased by 44% on average when dark treatments were used to relax NPQ, though, in contrast to the findings from the small lakes, there was little effect on CDOM estimates. We observed a statistically-significant, negative linear relationship between the photon flux density of in situ irradiance and the accuracy of taxonomic assignment by FP in Georgian Bay. Not discounting the correlations between light intensity and the accuracy of the FP that were observed in this study, we conclude that the applicability of the reference spectra to the system under investigation is a more important consideration than variability in natural irradiance conditions.

Effects of ultraviolet radiation on the productivity and composition of freshwater phytoplankton communities.

The net influence of ultraviolet radiation (UVR; 280-400 nm) on freshwater phytoplankton communities depends on the photon flux density, duration, and spectral quality of exposure and the UVR sensitivity of the assemblage in terms of photosynthetic impairment, biochemical composition, and nutrient assimilation mechanisms. Such effects are mitigated to varying degrees by photoacclimation and selective adaptation at the community level. Variation in UVR penetration among lakes is considerable, largely due to differences in chromophoric dissolved organic matter concentrations. Documented losses of areal daily primary production in lakes due to UVR range from negligible (2.5%) to appreciable (26%). UVR has the potential to alter algal biochemical composition and therefore indirectly affect higher trophic levels. There is evidence that algal nutritional status can influence UVR sensitivity, and that UVR can inhibit uptake and assimilation of inorganic nutrients, but results have been inconsistent. Taxonomic variability in susceptibility to the effects of UVR exists, and likely reflects variation in cell size and shape, concentrations of photoprotective pigments, and capacity to repair UVR photodamage. Suggestions for future research include: (1) resolution of taxon-specific UVR responses by way of single-cell techniques (e.g. enzyme-labelled fluorescence assays, microscope-based variable fluorometers) and (2) systematic comparative studies to link UVR exposure in natural habitats to community responses using the biological weighting function modelling approach. A more robust understanding of how sensitivity to UVR varies according to taxon and habitat is needed if predictions of its role in ecosystem functioning, particularly in connection with climate change, are to be meaningful.