Salinity predicts the distribution of chlorophyll a spring peak in the southern North Sea continental waters

In the North Sea, the coastal waters of Belgium and The Netherlands regularly exhibit intense spring phytoplankton blooms where species such as Phaeocystis recurrently form a potential ecological nuisance. In the Belgian and Dutch continental shelves (BCS and DCS), we observe a direct correlation between the chlorophyll a spring maximum (Chlmax) and the nutrients (DIN and DIP) available for the bloom. As the nutrients are themselves strongly correlated with salinity, a rationale is developed to predict Chlmax from winter salinity. The proposed rationale is first tested in a theoretical case with a 3D-biogeochemical model (3D-MIRO&CO). The method is then applied to independent sets of in situ observations over 20 years in the BCS and the DCS, and to continuous FerryBox data in April 2008. Linear regressions explain the relationships between winter nutrients and winter salinity (R-2 = 0.88 to 0.97 with model results, and R-2 = 0.83 to 0.96 with in situ data). The relationship between Chlmax and the available nutrients across the salinity gradient is also explained by yearly linear regressions (R-2 = 0.82 to 0.94 with model results, and R-2 = 0.46 to 0.98 with in situ data). Empirical DIP requirement' and 'DIN requirement' for the spring biomass bloom formation are derived from the latter relationships. They depend i.a. on the losses from phytoplankton during the spring bloom formation, and therefore show some interannual variability (8-12% for DIP and 13-20% for DIN). The ratio between nutrient requirements allows predicting in winter which nutrient will eventually limit the spring biomass bloom along the salinity gradient. DIP will generally be limiting in the coastal zone, whereas DIN will generally be limiting offshore, the switch occurring typically at salinity 33.5 in the BCS and 33.6 in the DCS. N reduction should be prioritized to limit Phaeocystis in the coastal zone, with target winter DIN:DIP ratios below 34.4 moIN moIP(-1) in the BCS, or 28.6 moIN molP(-1) in the DCS. (C) 2015 The Authors. Published by Elsevier B.V.