Algal assemblages have been widely used as an ecological indicator of aquatic ecosystem health conditions because of their specific sensitivity to a wide variety of environmental conditions. In turbid rivers, as in other aquatic systems, phytoplankton structure plays an important role in structuring aquatic food webs. Worldwide, phytoplankton is less studied in turbid, large tropical rivers compared to temperate river systems. The present study aimed to describe the phytoplankton diversity and abundance in a turbid tropical river (the Red River, northern part of Vietnam from 20°00 to 25°30 North; from 100°00 to 107°10 East) and to determine the importance of a series of environmental variables in controlling the phytoplankton community composition. Phytoplankton community was composed of 169 phytoplankton taxa from six algal groups including Bacillariophyceae, Chlorophyceae, Cryptophyceae, Euglenophyceae, Dinophyceae and Cyanobacteria. Community composition varied both spatially and with season. Sixteen measurement environmental variables were used as input variables for a three-layer backpropagation neural network that was developed to predict the phytoplankton abundance. Phytoplankton abundance was successfully predicted using the tagsig transfer function and the Levenberg-Marquardt backpropagation algorithm. The network was trained to provide a good overall linear fit to the total data set with a slope (R) and mean square error (MSE) of 0.808 and 0.0107, respectively. The sensitivity analysis and neutral interpretation diagram revealed that total phosphorus (TP), flow discharge, water temperature and P-PO43− were the significant variables. The results showed that the developed ANN model was able to simulate phytoplankton abundance in the Red River. These findings can help for gaining insight into and the relationship between phytoplankton and environmental factors in this complex, turbid, tropical river.
Peer-reviewed publication