The Fluorescence Analytical Technology For The Planktonic Algae In Major Lakes Of China

The harmful algal blooms (HABs) are becoming serious in lakes around theworld especially in China, which suggesting a need to develop a rapid and precisetechnology to monitor the composition of planktonic algae in situ. The3Dfluorescence spectra could could give the characteristic information of pigmentscomposition specific to algae species. In tis paper, wavelets analysis, multiwaveletsanalysis and Bayesian analysis were applied applied to extract feature spectra fromalgae3D fluorescence spectra, then, relevant reference spectra were established viahierarchical cluster analysis.finally, the fluorescence analytical technology weredeveloped to classify the planktonic algae in freshwater body combining withnon-negative least squares.On the basis of understanding the composition of planktonic algae in major lakesof China,27algae species (belong to22genus of6division) were cultured withdifferent experimental conditions, and three-dimensional fluorescence spectra wereobtained to develop the technology, the ultimate purpose is to determinate thecomposition of planktonic algae (dominant and subdominant division) at divisionlevel and to discriminate the HAB algae at genus level. The main research contentsare as follows:(1) Summing up the composition and biomass of planktonic algae in major lakes(such as Chaohu, Dianchi lakes and Taihu), rivers and reservoirs of China: theplanktonic algae community was made up of different species most of which belongedto one or two divisions. When HABs broke out, the diversification would reducesignificantly, one or two species dominated. Most of the HABs occurred owing toexcessive growth of cyanophyta (such as microcystis and anabaena) in major lakes ofChina, and HABs were often caused by bacillariophyta (such as cyclotella,stephanodisus) in rivers and reservoirs.(2) The relative standard deviation (RSD) was utilized as an indicator to analysis the measurement precision of three-dimensional fluorescence spectra, the resultsshowed that the3D fluorescence spectra of the algae had a good precision. Byvariance analysis, the significant differences were showed among the3D fluorescencespectra of different algae classes.(3) Two wavelets (db7, coif2) and two multiwavelets (SA4, GHM) were selectedto decompose the fluorescence spectra of planktonic algae (all the spectra had beenpreprocessed) into different scale vectors and time-series vectors, the optimum featurespectra were obtained by Bayesian analysis, then relevant reference spectra wereestablished via hierarchical cluster analysis to feature spectra, just as mentioned above,four kinds of technology were developed by non-negative least squares. All of thetechnology were tested by discriminating one-algae samples and mixed samples(including simulated mixed samples and laboratory mixed samples). For the one-algaesamples, the correctly discriminating ratios (CDRs) were98.6%,98.2%,98.3%, and97.8%at division level and86.2%,86.0%,84.1%, and84.0%at genus level,respectively. For the laboratory mixed samples, the average CDRs of dominantdivision were97.5%,97.5%,97.8%, and87.1%, with the average relative content of67.1%,67.9%,62.7%and57.5%at division level, and the average CDRs of dominantspecies were79.6%、80.1%,73.1%,75.8%at genus level. Some species which oftencaused HABs in lakes or rivers could be correctly classified with CDRs over75%.The technology developed by db7and coif2were better than the othes, and thecomplementarities among them were analyzed, then the two-ranks reference spectradatabase were constructed at genus level, the complementary technologies couldimprove the correctly disscriminate ratios of several species.(4) In order to develop a covenient, energy conservation, and rapid fluorescenceanalyzer,18excitation wavelengths (400,415,430,435,450,470,480,490,500,515,535,555,575,590,600,620,630,650nm) were selected to constitute new3Ddiscrete fluorescence spectra using stepwise discriminant analysis, simultaneously, theavailability of monochrome LEDs and the characteristic point of original spectrarequested careful consideration. db7and coif2wavelet were selected to decomposethe spectra into different vectors, two discrete fluorescence analysis technology were developed according to the same way above. The CDRs of one-algae samples were98.2%,96.6%at division level,81.0%,80.9%at genus level respectively. The CDRsof dominant algae in laboratory mixed samples were97.5%,96.2%at division leveland81.7%,76.7%at genus level respectively, also the subdominant algae could beclassified with CDRs of90.7%and88.8%at division level. Taking this as a technicalsupport, a fluorescence analyzer for in situ composition detection of planktonic algaein freshwater body could be developed.In conclusion, the technology developed in this paper could be applied topre-existing fluorescence spectrophotometer, meanwhile, could provide scientific andtechnical support for developing the new fluorescence analyzer.