Limnol Oceanogr. 2021 Aug;66(8):2967-2978. doi: 10.1002/lno.11851. Epub 2021 May 31.
Using National Lakes Assessment data, we evaluated the influence of total N (TN), total P (TP), and other variables on lake chlorophyll-a concentrations. With simple linear regressions, high TN/TP samples biased predictions based on TN, and low TN/TP samples biased predictions based on TP. The bias problem was corrected, and correlation was improved, by splitting the dataset at the TN/TP ratio we estimated to be indicative of a balanced supply and developing separate regressions that predict chlorophyll-a based on TP, TN, dissolved inorganic N (DIN), dissolved organic carbon (DOC), non-algal light attenuation, depth, area, latitude, elevation, and conductivity. Both nutrients were excellent predictors, and non-algal light attenuation was the next most influential predictor. The regression analysis suggested that a potential for P only limitation (high TN/TP, 17% of samples) or N only limitation (low TN/TP, 14% of samples) can be inferred at the extremes of the TN/TP range. However, 69% of samples had an intermediate TN/TP ratio where it is difficult to infer anything about potential nutrient limitations (biomass could be N limited, P limited, N and P co-limited, or not limited by nutrients at all). Our results show that when developing phytoplankton response relationships using cross-lake datasets that span a wide range of trophic states, it is important to consider whether and how biomass is influenced by confounding factors – such as differences in the relative supply of N and P – so that biomass is not underestimated or overestimated, and nutrient criteria are not under-protective or over-protective.