Sensors for Monitoring Harmful Algae, Cyanobacteria and Their Toxins

18 relationships. With these types of assays, you can “always get an easier answer, you just don’t know how good it is” without implementing the proper QC/QA metrics. Comparison to “gold standards” may not always be the best approach in ground-truthing. For example, there can be user error and discrepancies among microscopists performing cell counts. Mouse bioassays, once used as the “gold standard” in toxin assays, are somewhat arbitrary and the community has been moving away from this technique in recent years due to ethical concerns. • Q3: What’s needed/feasible to expand on these approaches? Complexities in assay chemistries, engineering intricacies of platforms, and data output provide equally complex challenges for ground-truthing and QA/QC methodologies. Nevertheless, empirical approaches and identification of caveats have allowed the HAB community to continue to work towards providing valuable protocols for HAB detection. In choosing an assay/platform, it is important to focus on the application – for example, if all that is needed is accuracy and reliability at a given regulatory cutoff (e.g., X ug/100 g tissue) the accuracy and precision at concentrations well above or below this do not matter. In other words, the technology has to be “fit for purpose”. Further, research applications (or even process monitoring at a water treatment plant) are different than regulatory applications. New technologies relating to food safety, clinical test procedures, or drinking water analysis require much more rigorous QA/QC, validations and even multi-laboratory validation or comparison of “approved or official methods”. Following from this, we also need to define “operational technology”. This concept spans the ultimate of “24/7, always on, never breaks, redundancy” (typical of the weather service) to a more simplistic “demonstration of principle” or “proof of concept” in the early stages of technology development. Defining the need will in turn define the required financial support. Technology developers should clearly distinguish developmental study data and performance or QA data for the finished prototype. The group discussed the goal of the ACT effort, e.g., helping new technology along for which there may not exist a sizable market. For many manufacturers, there is not enough return on investment to recoup development costs. The group discussed the high probability of failure for tech startup companies and the concept that the goal of ACT should not be to promote technology (for a long time) that “should” die in the market place. Further, there is a need for increased communication between researchers and companies, especially for young techniques in development. There can be issues with hardware performance versus biological/environmental (regional variation), as well as taxon specific needs. Optical platforms, such as the IFCB, were discussed as an example. There are issues related to hardware (e.g. beads for IFCB) versus data standardization, a need for defining windows for different growth phases, and considerations for packaging effects for fluorescence. Again, the issue of lack of standards for many congeners (e.g. > 200 for microcystin) was raised. Relying on blooms for ground- truthing is inherently difficult because results can be dependent on sampling time. Yet, there are differences in real world spatial/environmental data compared to lab calibration and a question of the appropriate number of samples to collect for ground-truthing – is >30 replicate samples over a season an adequate comparison for new versus existing methods? Furthermore, data interpretation can be problematic when decoupled from data collection