Sensors for Monitoring Harmful Algae, Cyanobacteria and Their Toxins

21 of algal composition and cell health over time. Although relationships can be built, they are very likely to be site specific. These statistical models should incorporate uncertainty given the other benefits of the measurements (e.g. large spatial coverage and frequent return for satellite; high frequency, real-time and low cost for in situ fluorometers). • Q2: Are there ‘gold standards’ of reference? The discussion around ‘gold standards’ in HAB research has been ongoing. While utilized for validation of system performance, in the biological realm they often need their own ground-truthing. One accepted method is microscopy, however there can be variability across microscopists’ scoring of a split sample (Godhe et al. 2007). While having one person perform microscopy within a project/monitoring program is feasible, this consistency framework breaks down when comparing results across broad temporal and spatial scales. This is further complicated when genetic diversity within morphologically identical species is included. Accuracy can vary across taxa at the species or genus level (e.g. Alexandrium versus Pseudo-nitzschia ), thereby limiting a universal approach for developing ‘gold standards’. This accuracy ties into sampling practices, such as how to overcome the difficulties in species that are difficult to culture (e.g. Dinophysis ), or exhibit different growth in culture (e.g. Microcystis exists as flocculated clumps in environmental samples, but remains dominantly uni-cellular in culture). One group focused on ‘gold standards’ related to qPCR and ELISA, as these two methods are very common for species and toxin detection, respectively. Currently, there is no ‘gold standard’ consensus for extraction of material (genetic and toxin) and this is a critical yet unmet need. Extraction methods should be defined as operations requiring verification testing . Microcystin was given as an example – the accepted protocol for toxin extraction is a triplicate freeze-thaw cycle, however it is unknown whether this method results in 100% cell lysis. Marine toxin extraction methods were not discussed in detail, other than to state that multiple extraction methods are required depending on the organism, sample matrix, and the complexity and importance of toxin profiles Probe sets are not universally defined, and there was discussion about whether we are hitting the right genetic targets. Researchers and kit vendors have validated probe sets, but currently there is no agreement on universal loci, at least for the microcystins (MCY-D, MCY-A, MCY-E). Furthermore, our knowledge of toxin genes is limited for most toxic HAB species, with the only other target known (outside those for microcystin), is for saxitoxin A (produced by Alexandrium spp.). For other species, there may be a suite of genes involved, further complicating the development of standards. Once validated, these toxin gene standards should be provided as part of vendor kits. For qPCR and SHA, cell counts are used to develop and validate standard curves, however we need to think about the reliability of using live cells. There was a suggestion to move towards the use of synthetic DNA molecules. Cell counts and extracted pigment concentrations are probably the realistic “gold standard” (although toxin concentration would be ideal if suitable models could be built). For future ACT testing of toxin kits, there will ideally be a set extraction method that can be used for verification/comparison. Even if the method is not perfect or the best, a consistently applied method is needed. Verification testing methods for immunoassays such as ELISA are generally well understood, covering topics such as recovery, precision, shelf-life/stability, and interfering