Heterotrophic nanoflagellates (HNF) and bacteria forming a tight predator-prey relationship in aquatic ecosystems are considered as key components of the microbial food webs. In this study, seasonal variation of pico- and nanoplankton community including HNF and bacteria was first investigated together with environmental variables during the period of 1993-1995 in Onagawa Bay on the northeastern Pacific coast of Japan. Based on the data obtained, a new scheme to describe seasonal fluctuations of the tight predator-prey relationship established between HNF and bacteria in nature is proposed. To confirm this scheme, an intensive monitoring at 1- to 3-day intervals and the measurement of bacterial growth rate were further conducted during the period of 1996-1997.

Results of the 1993-1995 investigations demonstrated that Onagawa Bay is of the typical embayments in temperate zone with a clear seasonal variability. Bacterial abundance varied mostly within a narrow range (500,000-2,000,000 cells/ml) throughout the year, while HNF abundance showed marked seasonal changes (30-6,700 cells/ml). Short-term observations at 2-day intervals made in July 1994 and in February/March 1996 revealed that peaks of bacterial abundance were usually followed by increases of HNF abundance with a lag of 2-6 days and both abundances changed cyclically with a 4-10 day period, indicating so-called predator-prey oscillations. This means that HNF quickly responds to increases in abundance and/or growth rate of bacteria. These predator-prey oscillations appeared as circular orbits, or 'predator-prey eddies', in phase space for short period e.g. a month. Location and magnitude of this HNF-bacterial eddy in phase space shifted with seasons. Therefore, the annual cycle of the HNF-bacteria relationship is supposed to appear as a continual migration of the predator-prey eddy in phase space. Since many previous works have also confirmed that a closely coupled relationship can be seen between HNF and bacteria, followings can be assumed: (1) predator-prey eddy of HNF-bacterial association proposed in this study is ubiquitous, (2) position of the eddy in phase space in particular season is determined by temporal environmental conditions, (3) extension of the annual migration of the eddy in phase space is determined by annual range of local environmental conditions particular to each ecosystem. The last means that each ecosystem has its particular 'attractor' that attracts the eddy not to protrude widely from a particular region in phase space.

Growth rate of bacteria fluctuated markedly with seasons, despite subtle seasonality in abundance. This contradictory phenomenon among marked seasonal variability of growth rate and little variability of abundance should be explained by top-down control by predators which can quickly remove increasing bacteria. Abundance of HNF actually varied widely with seasons, being higher in warm seasons. Abundances of HNF and bacteria fluctuated seasonally with a certain phase lag, forming the typical predator-prey oscillations. The phase lag was 2-7 days and the period was 2-16 days. No consistent seasonal change in both lag and period was observed throughout the year. These values of the lag and the period observed in Onagawa Bay are similar to those reported from other ecosystems. Therefore, it is considered that a similar relationship between them is commonly and firmly established over wide geographical ranges.