1. Introduction
Worldwide every year, fisheries, aquaculture, human health and tourism are threatened by harmful algal blooms (HABs) in marine, brackish, as well as continental waters. Although most phytoplankton species are benign, about 2% of them can cause harm through the production of toxins or by an excessive accumulated biomass, which can affect co-occurring organisms and alter food-web dynamics [1,2]. In addition to the ecological and economic damages, public health is also at risk: the consumption of shellfish that have fed on toxic phytoplankton and accumulated toxins, and exposure to the aerosols of HAB toxins can cause illness or even mortality. Depending on the species, it can take only a few toxic cells per liter to poison shellfish and make them unsuitable for human consumption [3]. Monitoring of microalgae is therefore required by all countries with a marine coastline. HAB monitoring programs are currently based on cells identified and counted by light microscopy and on the mouse bioassay for detecting biotoxins. The mouse bioassay for the detection of phytoplankton toxins in shellfish has recently been banned by the European Commission (July 2011), and there is a mandatory replacement by chemical methods (Liquid Chromatography-Mass Spectrometry, LC-MS) in the next three years.
However, the effectiveness of monitoring programs using light microscopic identification is limited by the fact that it is time consuming and that morphology, as determined by light microscopy, may be insufficient to give definitive species and toxin attribution. Thus, there is a need to implement molecular methods to ensure a fast and reliable species identification. Monitoring for toxic species using molecular techniques advances the state of knowledge for detection of harmful species because more samples can be analyzed in a shorter time period and with greater accuracy. Within the actual context of the dramatic decreasing number of taxonomic experts of phytoplankton [4], which are, notwithstanding, essential for other ecological studies, such techniques offer important advances. This is of particular interest for potentially toxic algae, because the difficulty in determining their exact identification by using light microscopy can have disastrous consequences for human health. Microarrays offer a near real-time ecosystem analysis and offer broader ecological interpretation of how key species, such as toxic algae, can extend their geographical distribution with climate change or can become invasive after introduction from remote areas [5]. Microarrays offer the most expeditious method to have high sample throughput with highly accurate species detection, in a universal approach [6,7,8]. In the FP7 EU project, MIDTAL (Microarrays for the Detection of Toxic Algae), an earlier protocol for detection of toxigenic microalgae by Gescher et al. [9] was optimized. Microarrays (as phylochips) detect multiple species simultaneously using species-specific probes that have been applied primarily for the detection of bacteria [10,11,12,13]. At present, 140 probes for various toxic algal species at various taxonomic levels are spotted onto the current generation of the MIDTAL microarray. As part of the MIDTAL project, the primary goal was to be able to infer cell numbers from the molecular signal to provide an early warning system for toxic algae. Because the MIDTAL microarray is a universal array that can be used globally, it offers a real possibility of detecting invasive species, especially in view of global warming where warm temperate species are moving northward, e.g., Gamberiodiscus.
In this study, we show the effectiveness of using microarrays for the detection of toxic algae and its combination with toxin detection. We compare these results with light microscopy data from a regular French monitoring network of toxic phytoplankton. The microarray used in this study represents the third generation array developed within the EU-MIDTAL project. In generation one, probes (18–22 nt) developed for Fluorescent in situ Hybridization (FISH) were used directly; in generation two, these FISH probes, and any newly designed probes, were lengthened to 25 or more nt; in generation three, an additional poly-T spacer to lift the probes farther above the surface was tested and optimized (Figure 1). At each generation, minor changes in the hybridization protocol were made and a final optimized protocol can be found in Lewis et al. [14].
Figure 1  Scheme of the development of the MIDTAL microarray. The scheme pictures the different microarray generations with its different probes, tests and enhancements of protocols (RNA and hybridization). (* Higher temperature during 3rd washing step).

2