Chapter 10 . Aigal Species Succession in Rivers
205
Chon et al. (2000), and Jeong et al. (2001a) previously documented the
predictability of ANN for species abundance and succession of freshwater algae
and macro-invertebrates. In this study, the Time-Delayed Recurrent Neural
Network (TDRNN) recognized distinct seasonal abundance and succession of
phytoplankton species as typical for the lower Nakdong River.
10.4.3
Elucidation of Ecological Hypotheses
Based on the Most Influencing Parameter sensitivity analysis (MIP) of the
validated RNN, both phytoplankton species were influenced largely by water
temperature and pH (Fig. 1O.3c). Dissolved Inorganic Nitrogen (DIN) was
important for the biovolume changes of S. hantzschii, but had relatively minor
effects on M. aeruginosa. Meteorological events, hydrological regimes, and
zooplankton abundances had less impact on dynamics of the phytoplankton. The
patterns observed in the phytoplankton are consistent with the conclusion of Joo et
al. (1997), that the lower Nakdong is a reservoir-like ecosystem. Phytoplankton
also is influenced during a short period in the summer rainy season, when there is
a sudden increase of discharge (Ha 1999), and a base flow that continues from fall
to late spring (Park 1998).
The occurrence of both bloom-forming species was previously found to be
correlated with increased pH, water temperature, and nutrient availability in lakes
and reservoirs (Reynolds 1984; Harris 1986; Sommer et al. 1986; Shapiro 1990).
Results of the RNN-based sensitivity analysis correspond with these previous
findings and indicate that short-term dynamics of bloom-formation in riverreservoir systems like the Nakdong River are mainly driven by physical-chemical
parameters such as temperature and pH, while long-term trends are basically
determined by the hydrological regime.
Sensitivity on Wide-ranged Disturbance (SWD) for the input variables revealed
information on relationships between input variables and both species (Fig. 10.4).
The response differed among the two species considered. Irradiance, evaporation,
Secchi depth, DO, conductivity, nitrate, phosphate, and dissolved silica
concentration influenced the increase or decrease of phytoplankton, while rainfall,
discharge, turbidity, ammonia concentration, and zooplankton abundances did not
have any recognizable effects. As we changed the range of disturbance for
evaporation and water temperature, both species exhibited dynamic variations.
205
Chon et al. (2000), and Jeong et al. (2001a) previously documented the
predictability of ANN for species abundance and succession of freshwater algae
and macro-invertebrates. In this study, the Time-Delayed Recurrent Neural
Network (TDRNN) recognized distinct seasonal abundance and succession of
phytoplankton species as typical for the lower Nakdong River.
10.4.3
Elucidation of Ecological Hypotheses
Based on the Most Influencing Parameter sensitivity analysis (MIP) of the
validated RNN, both phytoplankton species were influenced largely by water
temperature and pH (Fig. 1O.3c). Dissolved Inorganic Nitrogen (DIN) was
important for the biovolume changes of S. hantzschii, but had relatively minor
effects on M. aeruginosa. Meteorological events, hydrological regimes, and
zooplankton abundances had less impact on dynamics of the phytoplankton. The
patterns observed in the phytoplankton are consistent with the conclusion of Joo et
al. (1997), that the lower Nakdong is a reservoir-like ecosystem. Phytoplankton
also is influenced during a short period in the summer rainy season, when there is
a sudden increase of discharge (Ha 1999), and a base flow that continues from fall
to late spring (Park 1998).
The occurrence of both bloom-forming species was previously found to be
correlated with increased pH, water temperature, and nutrient availability in lakes
and reservoirs (Reynolds 1984; Harris 1986; Sommer et al. 1986; Shapiro 1990).
Results of the RNN-based sensitivity analysis correspond with these previous
findings and indicate that short-term dynamics of bloom-formation in riverreservoir systems like the Nakdong River are mainly driven by physical-chemical
parameters such as temperature and pH, while long-term trends are basically
determined by the hydrological regime.
Sensitivity on Wide-ranged Disturbance (SWD) for the input variables revealed
information on relationships between input variables and both species (Fig. 10.4).
The response differed among the two species considered. Irradiance, evaporation,
Secchi depth, DO, conductivity, nitrate, phosphate, and dissolved silica
concentration influenced the increase or decrease of phytoplankton, while rainfall,
discharge, turbidity, ammonia concentration, and zooplankton abundances did not
have any recognizable effects. As we changed the range of disturbance for
evaporation and water temperature, both species exhibited dynamic variations.