Supplementary file 1_Long-term trends and drivers of water color in Missouri reservoirs.docx

Contrasting water quality trends are occurring within and across North America, with waterbodies experiencing increasing phytoplankton blooms, increasing dissolved organic matter, or both, while others are becoming clearer and bluer; dramatically changing water color. To assess the spatial and temporal variability in water color, we quantified trends in satellite-derived dominant wavelength (λd) from 1984 to 2020 for 484 reservoirs across the state of Missouri using the LimnoSat-US dataset. Currently, the vast majority of Missouri reservoirs are classified as green and within a range (538–555 nm) that lies closer to the brown color endmember. Nearly one-third of reservoirs (n = 159) experienced significant temporal shifts in water color, with more (n = 91) negative (e.g., bluer) than positive (n = 68) λd trends. Linear mixed-effect models indicate that periods of extreme wetness and drought are associated with browner and bluer waters, respectively, and boosted regression trees further reveal that waterbody and watershed characteristics are important predictors for water color trends. We also analyzed trends in summer water quality (WQ) parameters from two long-term monitoring programs to evaluate independent and synchronous changes with λd. We provide analyses showing that particulate inorganic matter and Secchi depth most strongly correlated with λd, and total nitrogen and total phosphorus concentrations that are not typically associated with satellite-derived data have greater co-variance with λd than chlorophyll a. Although bluer waters often reflect reductions in inorganic turbidity, our findings show that they can at times coincide with increases in chlorophyll a. We further demonstrate that while λd trends broadly align with changes in water quality, co-occurring water quality and color trends in Missouri reservoirs at times defy a simplistic canonical interpretation, particularly in eutrophic waterbodies where changes in nutrient concentrations, chlorophyll a and water color can occur independent of each other. Our results help explain some of the previously observed heterogeneous controls on water color and emphasize the importance of integrating water quality data alongside commonly used landscape and morphological features.