A comparison of primary production models in an area of high mesoscale variability (South Shetland Islands, Antarctica)

Three types of primary productivity (PP) models were evaluated in a mesoscale area around the South Shetland Islands (Antarctica). Input variables were: phytoplankton carbon biomass, Chlorophyll a, sea water temperature, daily irradiance, among others, collected in situ during an oceanographic cruise (COUPLING, January 2010). Models of the first type were based on Chl a measurements: the widely used model VGPM (Behrenfeld and Falkowski, 1997) and a derived version developed for the Western Antarctic Peninsula (Dierssen et al., 2000). The second type included two models based on phytoplankton carbon biomass: one developed for the whole Southern Ocean (Arrigo et al., 2008) and one based on the Metabolic Theory of Ecology developed by López-Urrutia et al. (2006), being the first time that a model with these features is used for Antarctic waters. The third type was an updated version of the carbon-based model CbPM (first described by Behrenfeld et al. (2005)) based on the Chl a/carbon biomass ratio modulation. The degree of agreement among the results between the different types of models turned out to be low (> 30% of difference), but high within models of the same type (< 10% of difference). Biomass-based model predictions differed the most from those estimated by the other two types. The differences in PP estimates were primarily attributed to the different ways these models treat the phytoplankton assemblage, along with the difference in input variables. Among the five models evaluated, the output from the modified version of the CbPM showed the lowest bias (0.55) being the most realistic. It made a special attempt to detect the factors controlling phytoplankton physiological state, showing a nutrient limitation towards the Drake area similar to the one observed for the in situ PP values.