The effects of tropical cyclones on seasonal variations in phytoplankton biomass were studied at a station on the northern coast of Taiwan. Water temperature at the study site vaned between 17 and 30° in 1994 Both chlorophyll a concentration and Synechococcus abundance were low in the winter and started to increase when surface water became warmer than 25OC In July and August, 4 cyclones struck the study site All cyclones dramatically reduced phytoplankton biomass in the water column, and caused significant increases in nitrate, nitrite, phosphate, and silicate concentrations Blooms of phytoplankton with total chlorophyll a concentration as high as 4.2 pg 1 ' were observed after cyclonic disturbances, and were mainly composed of species greater than 5 W I I in size Our results sugqest that cyclones disturb coastal marine ecosystems in a consistent way, and induce ordered and predictable changes in a phytoplankton community.

EXECUTIVE SUMMARYCoastal environments are an important component of the global carbon cycle, and probably more vulnerable than the open ocean to anthropogenic forcings. Due to strong spatial heterogeneity and temporal variability, carbon flows in coastal environments are poorly constrained. Hence, an integrated, international, and interdisciplinary program of ship-based hydrography, Voluntary Observing Ship (VOS) lines, time-series moorings, floats, gliders, and autonomous surface vessels with sensors for pCO 2 and ancillary variables are recommended to better understand present day carbon cycle dynamics, quantify air-sea CO 2 fluxes, and determine future long-term trends of CO 2 in response to global change forcings (changes in river inputs, in the hydrological cycle, in circulation, sea-ice retreat, expanding oxygen minimum zones, ocean acidification, …) in the coastal oceans. Integration at the international level is also required for data archiving, management, and synthesis that will require multi-scale approaches including the development of biogeochemical models and use of remotely sensed parameters. The total cost of these observational efforts is estimated at about 50 million US dollars per year.