Temperature influence on phytoplankton community growth rates

A large database of field estimates of phytoplankton community growth rates in natural populations was compiled and analyzed to determine the apparent temperature effect on phytoplankton community growth rate. We conducted an ordinary least squares regression to optimize the parameters in two commonly used growth‐temperature relations (Arrhenius and Q10 models). Both equations fit the observational data equally with the optimized parameter values. The optimum apparent Q10 value was 1.47 ± 0.08 (95% confidence interval, CI). Microzooplankton grazing rates closely matched the temperature trends for phytoplankton growth. This likely reflects a dynamic adjustment of biomass and grazing rates by the microzooplankton to match their available food source, illustrating tight coupling of phytoplankton growth and microzooplankton grazing rates. The field‐measured temperature effect and growth rates were compared with estimates from the satellite Carbon‐based Productivity Model (CbPM) and three Earth System Models (ESMs), with model output extracted at the same month and sampling locations as the observations. The optimized, apparent Q10 value calculated for the CbPM was 1.51, with overestimation of growth rates. The apparent Q10 value in the Community Earth System Model (V1.0) was 1.65, with modest underestimation of growth rates. The GFDL‐ESM2M and GFDL‐ESM2G models produced apparent Q10 values of 1.52 and 1.39, respectively. Models with an apparent Q10 that is significantly greater than ~1.5 will overestimate the phytoplankton community growth response to the ongoing climate warming and will have spatial biases in estimated growth rates for the current era. Key Points Estimations of Q10 and Arrhenius equation parameters were made using field‐measured phytoplankton community growth rates The Arrhenius and Q10 equations do an equally good job of estimating the temperature dependence of phytoplankton community growth rates The optimal apparent Q10 value is 1.5. Models should capture a community growth‐temperature response equal to this value to avoid bias