Toxic and non-toxic Nodularia strains can be distinguished from each other and from eukaryotic algae with chlorophyll fluorescence fingerprinting Mika Kera ¨nen a , Eva-Mari Aro a , Olli Nevalainen b , Esa Tyystja ¨rvi a, * a Laboratory of Plant Physiology and Molecular Biology, University of Turku, FI-20014 Turku, Finland b Department of Information Technology and TUCS (Turku Centre for Computer Science), University of Turku, FI-20014 Turku, Finland 1. Introduction Correct identification of microalgae and cyanobacteria species in natural waters is essential for environmental management and research. Understanding of population dynamics of algae and the monitoring of harmful algal and cyanobacterial species need methods for rapid classification of the species. Toxic cyanobacteria species are found globally, and the rapid detection of toxic species is important in drinking water management, particularly at the beginning of a cyanobacterial or algal bloom. Nodularia is one of the cyanobacteria genera that bloom annually in the Baltic Sea. Microalgae have traditionally been classified using morpholo- gical criteria, but microscopy is time-consuming and expensive. Several automatic methods have actually been developed for morphological identification. Computer vision has been used to identify phytoplankton cells and colonies (Uhlmann et al., 1978), dinoflagellates (Culverhouse et al., 1996), cyanobacteria (Walker and Kumagai, 2000) and diatoms (Fischer and Bunke, 2001) as well as zooplankton (Benfield et al., 1998). Flow cytometry has also been developed as an identification method (Dubelaar and Jonker, 2000). The chlorophyll a (chla) molecules of photosystem I (PS I) and photosystem II (PS II), located in the thylakoid membranes of chloroplasts, emit red fluorescence under visible or UV light. Both emission spectroscopy and fluorescence induction measurements have been widely used in photosynthesis research and stress physiology of plants, algae and cyanobacteria. The differences in pigment composition between cyanobacteria and between differ- ent classes of algae, as revealed by emission spectroscopy, form a basis for differentiating phytoplankton to ‘spectral groups’ (Beutler et al., 2002). However, fluorescence emission spectra do not have enough information for automatic species-level identification of freshwater algae (Hilton et al., 1989). The photosynthetic machinery of cyanobacteria contains two fluorescent components. Phycobilisomes emit mainly around 640 nm, and chla emission peaks at 685 nm. Contrary to fluorescence originating from phycobilisomes or other cell components, chla fluorescence is not stable but fluctuates in response to changes in the reduction state of the photosynthetic electron transfer chain. Thus, chla fluorescence yield is an internal probe of photosynthesis (for reviews, see Baker and Oxborough, 2004 and Wilhelm et al., 2004). At physiological temperatures, chla fluorescence originates mainly from PS II whereas PS I has a high fluorescence yield only at low temperatures (Goedheer, 1972; Itoh and Sugiura, 2004). Chla fluorescence is typically studied by first incubating the sample in the dark and then rapidly switching on the exciting light. This procedure causes a series of changes in chla fluorescence yield, and the resulting time series is called the fluorescence induction curve. The pulse amplitude modulation Harmful Algae 8 (2009) 817–822 ARTICLE INFO Article history: Received 14 October 2006 Received in revised form 11 November 2007 Accepted 4 December 2007 Keywords: Automatic species identification Chlorophyll fluorescence Cyanobacteria Neural network Nodularia ABSTRACT Chlorophyll fluorescence induction curves of toxic and non-toxic strains of the cyanobacterium Nodularia were measured and compared with fluorescence curves measured from four species of eukaryotic algae. Both cyanobacteria and algae were isolated from the Baltic Sea. The results show that Nodularia strains can be distinguished from the eukaryotes by applying a pattern recognition procedure to the fluorescence induction curves, suggesting that the fluorescence fingerprinting technique might be useful in environmental monitoring of marine algae. The six studied Nodularia strains could not be distinguished from each other from their fluorescence induction kinetics. However, their fluorescence curves fell into two clear categories, the toxic and the non-toxic Nodularia. Emission spectroscopy and differences in the fluorescence induction curves showed that the ratio of the intensity of the Photosystem I emission peak to the Photosystem II peak is higher in non-toxic Nodularia than in the toxic strains, suggesting that the toxicity affects the structure of the photosynthesis machinery. The effect on photosynthesis may be related to the ability of the microcystins to chelate iron. ß 2009 Elsevier B.V. All rights reserved. * Corresponding author. Fax: +358 2 333 8075. E-mail address: esatyy@utu.fi (E. Tyystja ¨ rvi). Contents lists available at ScienceDirect Harmful Algae journal homepage: www.elsevier.com/locate/hal 1568-9883/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.hal.2007.12.023