208
K.-S. Jeong . F. Recknagel . G.-J. Joo
10.4.3.2
Stephanodiscus hantzschii
Stephanodiscus hantzschii blooms in the lower Nakdong River could be explained
by hydrodynamic factors. The results of SWD indicated that water temperature,
Secchi depth, pH, and dissolved silica were strongly related to the dynamics of the
species (Fig. 10.4). Stephanodiscus is known to prefer low water temperatures,
and several European eutrophic rivers had blooms of this genus in the winter
(temperatures around 15°C) (Descy et al. 1987). Ha (1999) reported that S.
hantzschii blooms in the lower Nakdong River occurred at much lower
temperatures (4-8°C). In this SWD study, when the water temperature exceeded 78 oe, biovolume of S. hantzschii sharply decreased.
Because diatoms use dissolved silica to generate frustules, the concentration of
Si0 2 is an important determinant of their growth (Round et al. 1990). An enclosure
experiment conducted by Ha et al. (1998) and Ha and Joo (2000) indicated that S.
hantzschii was dominant at low Si0 2 (1-1.5 mg L1
), while Fragilaria crotonensis
and Synedra acus had a competitive advantage in Si0 2 -added enclosures. From the
field data, ANN could recognize the importance of dissolved silica, which was
similar to that indicated by the experimental results.
Stephanodiscus hantzschii abundance increased at pH levels around 7.5-8.5 and
started to decrease after pH 9.0. Similar to the case of M. aeruginosa, this situation
can be interpreted by the e0 2 -pH complex. Most algal species, except bluegreens, are sensitive to dissolved carbon dioxide in the water, because they are
unable to utilize the other dissolved forms that occur at higher pH. The decrease of
S. hantzschii at higher pH values could be explained by this phenomenon.
10.5
Implications of Ecological Informatics for Limnology
Natural ecosystems are distinctly non-linear, dynamic and complex. Powerful
mathematical and computational techniques are required to elucidate and predict
driving forces and processes underlying extreme ecosystem behaviors such as
algal bloom events (Straskraba 1994). As shown in this study, artificial neural
networks prove to be one suitable computational technique for these purposes.
However, the newly emerging discipline of ecological informatics provides a
variety of computational techniques such as fuzzy logic, cellular automata,
evolutionary algorithms and adaptive agents (e.g. Fielding 1999; Whigharn and
Recknagel 2001a; Whigham and Recknagel 2001b; Bobbin and Recknagel 2001;
Jeong et al. 2001b; Recknagel 2001). Paired with growing power of computers
these techniques extent, complement, reinforce or hybridise ecological modeling
techniques towards more realistic modelling of limnological phenomena at
different levels of organization and complexity (see Fig. 10.5).
Hybrid architectures of empirical models as suggested by Medsker (1996) may
further encourage inter-disciplinary research between ecology and computer
K.-S. Jeong . F. Recknagel . G.-J. Joo
10.4.3.2
Stephanodiscus hantzschii
Stephanodiscus hantzschii blooms in the lower Nakdong River could be explained
by hydrodynamic factors. The results of SWD indicated that water temperature,
Secchi depth, pH, and dissolved silica were strongly related to the dynamics of the
species (Fig. 10.4). Stephanodiscus is known to prefer low water temperatures,
and several European eutrophic rivers had blooms of this genus in the winter
(temperatures around 15°C) (Descy et al. 1987). Ha (1999) reported that S.
hantzschii blooms in the lower Nakdong River occurred at much lower
temperatures (4-8°C). In this SWD study, when the water temperature exceeded 78 oe, biovolume of S. hantzschii sharply decreased.
Because diatoms use dissolved silica to generate frustules, the concentration of
Si0 2 is an important determinant of their growth (Round et al. 1990). An enclosure
experiment conducted by Ha et al. (1998) and Ha and Joo (2000) indicated that S.
hantzschii was dominant at low Si0 2 (1-1.5 mg L1
), while Fragilaria crotonensis
and Synedra acus had a competitive advantage in Si0 2 -added enclosures. From the
field data, ANN could recognize the importance of dissolved silica, which was
similar to that indicated by the experimental results.
Stephanodiscus hantzschii abundance increased at pH levels around 7.5-8.5 and
started to decrease after pH 9.0. Similar to the case of M. aeruginosa, this situation
can be interpreted by the e0 2 -pH complex. Most algal species, except bluegreens, are sensitive to dissolved carbon dioxide in the water, because they are
unable to utilize the other dissolved forms that occur at higher pH. The decrease of
S. hantzschii at higher pH values could be explained by this phenomenon.
10.5
Implications of Ecological Informatics for Limnology
Natural ecosystems are distinctly non-linear, dynamic and complex. Powerful
mathematical and computational techniques are required to elucidate and predict
driving forces and processes underlying extreme ecosystem behaviors such as
algal bloom events (Straskraba 1994). As shown in this study, artificial neural
networks prove to be one suitable computational technique for these purposes.
However, the newly emerging discipline of ecological informatics provides a
variety of computational techniques such as fuzzy logic, cellular automata,
evolutionary algorithms and adaptive agents (e.g. Fielding 1999; Whigharn and
Recknagel 2001a; Whigham and Recknagel 2001b; Bobbin and Recknagel 2001;
Jeong et al. 2001b; Recknagel 2001). Paired with growing power of computers
these techniques extent, complement, reinforce or hybridise ecological modeling
techniques towards more realistic modelling of limnological phenomena at
different levels of organization and complexity (see Fig. 10.5).
Hybrid architectures of empirical models as suggested by Medsker (1996) may
further encourage inter-disciplinary research between ecology and computer