How far do physical factors determine phytoplankton biomass in lakes?

Abstract

A multivariate calibration model, MVC, based on seven physical parameters driving the ecosystem in four lakes was able to explain 38% to 61% of the variation in the seasonal phytoplankton observations (as Chl- a). These values can be compared to estimates which show that biological interactions (expressed as dynamic chaos trajectories) may explain about 50% of the variation in a marine diatom (phytoplankton) population. Thus, physical factors and biological interactions explain as a first approximation about 50% each of the variation in phytoplankton biomass. Visual comparison between observed and predicted phytoplankton time series showed best agreement for the deep (⪢ 100 m), oligotrophic Lake Mjøsa. For two shallow lakes the phytoplankton blooms were well predicted in some years, but were grossly underestimated in others. This suggests that there may be a dichotomy among the bloom-forming factors. Outliers in the MVC models could, in some cases, be related to extreme values of physical factors, but as often not. MVC predicted summer average and summer peak phytoplankton biomass better than the OECD equations normally used ( Eutrophication of Waters. Monitoring, Assessment, and Control, OECD, Paris, 1982). Thus, since for many lakes there now exist satisfactory calibration data sets, MVC models could replace OECD type models (which are based on data from many lakes) as a tool for predicting lake response to changing nutrient loads. However, in spite of our ability to predict the day of the vernal bloom from a published regression equation using annual temperature and summer mean Chl- a as regressors (observed versus predicted day gave r 2= 0.47, p<0.0001), we were not able to find any correlation between the observed bloom day and the bloom day of the MVC predicted time series ( p⪢0.05).