Modeling the effects of temperature, light, and nutrients on primary productivity: An empirical and a mechanistic approach compared

The prospect of reliably predicting production from environmental variables in heterogeneous estuarine environments offers clear logistic advantages over measurement with the 14C technique, particularly if detailed coverage (in space and time) is desired. To that end, two different techniques for predicting phytoplankton production were tested and compared using data from Narragansett Bay, R.I., and experimental estuarine ecosystems. Primary productivity was first estimated via an empirical technique (principal component regression), which incorporated temperature, light, and nutrients as independent variables. This approach was compared with a traditional mechanistic formulation which included the same environmental variables as the major limiting factors. Both models provided good fits to the data (explaining 77–82% of the total variance). The predictive ability of the mechanistic model was significantly greater (P < 0.05) than the principal component regression when tested on two independent data sets. Predictive ability between the two models was statistically compared by F#x2010;tests on the residual variability from each method.