. PO -P l and two levels of combined additions were followed for 3 days using 6 l experimental units. To evaluate the 4 nutrient limitation patterns, time series of chlorophyll a were analysed using polynomial regression models and ranking method, taking advantage of the relatively constant experimental error. Apparent nutrient depletion rates and ratios were estimated, and compared with the changes in particulate nutrient ratios. During the spring diatom bloom in 1995, ambient inorganic nutrient concentrations were still high, and thus phytoplankton biomass did not respond to additions of nutrients. Ž Ž . y1 y1 Ž Chlorophyll a specific nutrient depletion rates were low 0.01-0.12 mg N mg chl a h and 0.002-0.016 mg P mg chl . y1 y1 . a h and linear over time, thus also revealing that phytoplankton was not limited by these nutrients in that time. In June 1994, there was an areal shift from N limitation in the outer Irbe Strait towards co-limitation in the southern Gulf. Later in July 1994, during the bloom of N-fixing Aphanizomenon flos-aquae, the N limitation was obvious for the whole study Ž Ž . y1 y1 area. For this period chlorophyll a specific nutrient depletion rates were high 0.36-0.67 mg N mg chl a h and Ž . y1 y1 . 0.089-0.135 mg P mg chl a h , and added nutrients were almost totally depleted during the first light period. After the collapse of cyanobacterial bloom in August 1993, the experiment carried out in the southern Gulf indicated P limitation of phytoplankton. The central Gulf was obviously co-limited, while the area between northern Gulf and outer Irbe Strait was N-limited. Our results indicate that phytoplankton in the Gulf of Riga, earlier considered strictly as P-limited, is at least until late-summer period N-or co-limited. It seems also obvious that there exists a spatial tendency in the phytoplankton limitation patterns, generally from more P-or co-limited southern Gulf towards more N-limited northern basin. q