Premium
A consumer's guide to phytoplankton primary productivity models
Author(s) -
Behrenfeld Michael J.,
Falkowski Paul G.
Publication year - 1997
Publication title -
limnology and oceanography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.7
H-Index - 197
eISSN - 1939-5590
pISSN - 0024-3590
DOI - 10.4319/lo.1997.42.7.1479
Subject(s) - phytoplankton , irradiance , biomass (ecology) , primary productivity , productivity , photic zone , environmental science , photosynthesis , atmospheric sciences , primary producers , ecology , equating , primary production , variable (mathematics) , mathematics , biology , statistics , ecosystem , nutrient , botany , physics , mathematical analysis , macroeconomics , quantum mechanics , rasch model , economics
We describe a classification system for daily phytoplankton primary productivity models based on four implicit levels of mathematical integration. Depth‐integrated productivity models have appeared in the literature on average once every 2 years over the past four decades. All of these models can be related to a single formulation equating depth‐integrated primary production (∑ PP ) to surface phytoplankton biomass (C surf ), a photoadaptive variable ( P b opt ), euphotic depth (Z eu ), an irradiance‐dependent function ( F ), and daylength ( DL ). The primary difference between models is the description of F, yet we found that irradiance has a relatively minor effect on variability in ∑ PP. We also found that only a small fraction of variability in ∑ PP can be attributed to vertical variability in phytoplankton biomass or variability in the light‐limited slope for photosynthesis. Our results indicate that (1) differences between or within any model category have the potential to improve estimates of ∑ PP by <10%, so long as equivalent parameterizations are used for C surf and P b opt , and (2) differences in estimates of global annual primary production are due almost entirely to differences in input biomass fields and estimates of the photoadaptive variable, P b opt , not to fundamental differences between model constructs.