At the beginning of the 21st century the blooms of the toxic benthic marine dinoflagellate genus Ostreopsis Schmidt appeared as an emerging potential threat to human health in Mediterranean coastal areas. This microalga of tropical origin that produces palytoxin (PLTX; Rossi et al. 2010) and associated analogues has been blooming recurrently at some locations in the Mediterranean during the last 20 years (e.g. Vila et al. 2001, Aligizaki and Nikolaidis 2006, Mangialajo et al. 2011). Its presence was previously detected in tropical and temperate coasts of Australia, the Indian Ocean, eastern Asian, the Caribbean sea, Brazil and, in the last decade, in colder waters of the northern Sea of Japan, New Zealand, Portugal, the Atlantic coasts of France and Spain (e.g. Selina and Orlova 2010, Rhodes 2011 and references therein). This apparent increased occurrence in both temperate and tropical regions (Rhodes 2011) has also been noticed with other harmful microalgae. It is not yet clear whether this is due, in part, to improved detection and monitoring capacity or whether the combination of global warming and anthropogenic forcing (in particular, linked to the alteration of coastal ecosystems due to excessive human use) is responsible.

The genus Ostreopsis grows well in shallow water as an epiphyte of macroalgae and dead corals, or covering sand, pebbles and rocks; it also occurs as plankton in the water column. It proliferates mainly attached to surfaces by producing extracellular polysaccharide mucilage derived from pusule fibrous material and mucocysts (see images in GEOHAB 2012 and Honsell et al. 2013). In temperate waters, Ostreopsis species bloom in the summer-autumn seasons (Shears and Ross 2009, Mangialajo et al. 2011) and produce a characteristic rusty-brown–coloured film over the colonized benthic substrate. In addition, high densities of cells swimming freely or embedded in the mucous matrix can be found in the water column and at the air-water interface as floating aggregates.

Ostreopsis proliferations are a serious concern because the PLTX-like group of biotoxins has been implicated in fatal seafood poisonings at tropical latitudes after ingestion of clupeid fishes or crabs (Alcala et al. 1988, Randall 2005) contaminated with bioaccumulated PLTX. To date, fortunately, no Ostreopsis-related food-borne illnesses have been reported in Europe. However, on certain Mediterranean beaches, the presence of the species Ostreopsis cf. ovata has been accompanied by respiratory symptoms and skin and mucosa irritation in humans (see references in Table 1). The affected people had spent some hours, during their recreational or working activities, in beach areas affected by Ostreopsis proliferations, or reside nearby. Acute symptoms include respiratory irritation (rhinorrhea, pharyngeal pain, dry or mildly productive cough, nose irritation), general malaise, headache, fever (≥38°C), eye irritation and/or dermatitis (Durando et al. 2007, Tichadou et al. 2010). Most of the symptoms disappeared within a few hours without specific medication when people moved away from the affected area in the vicinity of the Ostreopsis bloom; in only a few severe cases was hospitalization necessary (Durando et al. 2007).

It has been hypothesized that direct contact with seawater containing the microalga and/or with marine aerosols produced during its presence could be the plausible cause of the health disorders. However, the hypothesis has not been confirmed yet for several reasons. First, the profile of clinical manifestations is non-specific, with most cases corresponding to people who were sporadically exposed to marine aerosols and seawater. Hence, the recurrent Ostreopsis blooms are not always accompanied by concomitant symptoms that would generate health reports. This causes under-diagnosis and, thus, under-reporting to the surveillance health system (Tichadou et al. 2010, Pfannkuchen et al. 2012). Second, the estimations of the Ostreopsis concentrations in the affected area were usually conducted only after symptoms were reported. Often, the cell numbers found in plankton are highly variable over short periods (up to two orders of magnitude within few hours; Vila et al. 2008). Thus, the low cell abundances found in plankton on days following the health outbreak hampered our ability to establish a clear relationship between the Ostreopsis blooms and the problems experienced by humans during this study. Third, the nature of the irritative agents has not yet been identified. Although PLTX-like compounds have been suggested as potential agents, the few attempts to detect them in marine aerosols have been inconclusive. In the first study conducted on aerosol samples collected during the Ostreopsis cf. ovata blooms in Sant Andreu de Llavaneres (Catalan coast, NE Spain) in 2009 and 2010 (Casabianca et al. 2013), PLTX-like compounds were not detected, although Ostreopsis cells were identified by molecular tools. In the sampling performed along the Tuscan coast (Italy) in the same years (Ciminiello et al. 2014), a single positive result was obtained. However, this sample was taken in absence of O. cf. ovata in the water samples and no toxicological data on PLTXs or PLTX-like compounds by inhalation exposure were conducted in that study.

In the NW Mediterranean, various studies have been conducted on Sant Andreu de Llavaneres beach (41°33.13′N; 2°29.54′E), which is an Ostreopsis hot spot where annual blooms have been occurring in the last few decades (Mangialajo et al. 2011, Vila et al. 2012a). At this station, O. cf. ovata spans at least two months at high cell abundances (>10⁵ cells g^–1 of macroalgae fresh weight [FW]) (Vila et al. 2012b). This is very atypical, as blooms in other places along the NW Mediterranean Sea often last for 2 or 3 weeks. The first outbreak in 1998 was associated with benthic fauna mortality (Vila et al. 2008). In the summer of 2004, 74 cases of respiratory disorders were recorded (with up to 200 people potentially affected), coinciding with the presence of the microalgae (Àlvarez et al. 2005, Vila et al. 2008; Table 1). In this case, the causal relationship between symptoms and O. cf. ovata was established by chance, because the health problems occurred few days after phytoplankton samples had been taken in the framework of a wider programme "Monitoring of Harmful Algal Blooms in Catalonia”, a contract between ICM-CSIC and the Water Catalan Agency (Generalitat de Catalunya) that controls harmful algal blooms (HABs) along the Catalan coast. Later, in 2006, a new health outbreak affected 37 people in Sant Andreu de Llavaneres (Àlvarez 2006; Table 1). The association with the presence of the microalgae was uncertain, because seawater samples were taken two days after the respiratory irritation was reported and O. cf. ovata cell densities in plankton were low (in the order of 2×10³ cells L^–1). Subsequent respiratory symptoms, mucosa irritation and wound infections were detected in the population inhabiting, working and/or swimming in this beach area in the summer period but had not been reported. Thus, the direct link with the O. cf. ovata blooms remains unclear due to the temporal mismatch between the ecological and the epidemiology studies, as noted earlier (Tichadou et al. 2010, Pfannkuchen et al. 2012).

To clarify this uncertainty, we designed an integrated follow-up epidemiology study in a human population exposed to an Ostreopsis bloom in parallel to the field monitoring of the bloom event. While this study focuses only on respiratory disorders, the outcomes of this research can contribute to the assessment of the general risks to the public health and the environment posed by Ostreopsis outbreaks.

Our work constitutes the first monitoring of an Ostreopsis bloom conducted simultaneously and in coordination with prospective epidemiological follow-up research. The main strengths of our study are that 1) it was conducted on a well-defined human cohort, 2) the cohort followed was permanently exposed to the environmental conditions related to a long-lasting Ostreopsis bloom; and 3) the cohort was monitored for six months, i.e. before, during and after the bloom. Although a larger human cohort would have been epidemiologically suitable, our study fulfilled these three important conditions, which are very difficult to achieve in field studies.

Our results support the existence of a causal link between blooms of this toxic dinoflagellate and certain health impacts on humans. Paradoxically, the health affectation in our study site occurred during only a few days, whereas the Ostreopsis cell proliferation persisted for more than two months. This finding could result from the interaction of complex mechanisms involving meteorological and oceanic conditions, the physiology of Ostreopsis cells and the particular sensitivity of the affected individuals. Our working hypotheses for future studies are discussed next.

In 2013, the Ostreopsis bloom exhibited the typical seasonal pattern described earlier in Sant Andreu de Llavaneres (Mangialajo et al. 2011, Vila et al. 2012a, b). The meteorology conditions during the summer-autumn season fitted within the general climatology described in the area. Cells bloomed exponentially after approximately a two-week period, with temperatures above 20°C and relatively calm weather (based on the recorded wind intensities). Overall, cell abundances were in the range reported elsewhere in the area (e.g. Mangialajo et al. 2011, Vila et al. 2012a, b). The epiphytic Ostreopsis cf. ovata population remained for more than two months at high concentrations and within the same order of magnitude (10⁵ cells g^–1 FW). In the water, the cell numbers followed a similar dynamics, although with higher variability (within two orders of magnitude, from 10⁴ to 5×10⁵ cells L^–1). As described by Vila et al. (2008), high daily variations in the planktonic cell numbers in the water column could be partly due to biological rhythms of cell detachment from the macroalgae. Furthermore, water motion controlled by the major hydrodynamic patterns near the Catalan coast (see e.g. Sabatés et al. 2001, and references there in), combined with the daily breeze regime and swell, would then cause the dilution or concentration of the Ostreopsis planktonic population. As for the epidemiological study in this Ostreopsis hot spot, the observed human cohort suffered a similar attack rate (81% of the studied cohort) to that of the previous reported outbreaks (Àlvarez et al. 2005, Àlvarez 2006, Vila et al. 2008). However, in 2013 the affected people experienced fewer symptoms and of milder intensity than in 2004 and 2006. At present, there is a lack of similar epidemiology studies conducted simultaneously with the ecological monitoring. This lack of data hampers our ability to hypothesize about the factors involved in the interannual variation of the symptoms of human health, so further research is required. The interannual variability of both Ostreopsis bloom intensity and other meteorological and oceanographic conditions may also be at play and will be considered elsewhere.

In the 2013 Ostreopsis bloom that we studied, the health symptoms occurred mainly during a particularly short time window of the event, coinciding with the transition from the exponential to the stationary phase of the Ostreopsis population, with high planktonic and epiphytic cell concentrations. This finding suggests that Ostreopsis cells may produce certain toxic or allergenic substances responsible for the observed symptoms during a specific physiological state coinciding with a specific phase of the bloom. As has been observed in toxin producing microalgae, toxin cell concentrations may vary with the physiological phase as well as in response to environmental forcings (e.g. Bates et al. 1991, Pan et al. 1996, Taroncher-Oldenberg et al. 1997), although the response can be species-specific. For instance, in laboratory experiments (Bolli et al. 2007), paralytic shellfish toxin cell content in Alexandrium catenella and A. minutum decreased from the exponential to the stationary phase, and it was also lower under turbulent than calm conditions.

During the stationary phase of the bloom, in spite of high epiphytic cell numbers until the bloom collapsed, the health impacts were negligible. Only the person affected by allergy exhibited two symptoms potentially related to the Ostreopsis bloom in October (Table 2). In the case of respiratory problems associated with the Karenia brevis blooms in Florida and the Gulf of Mexico, the highest sensitivity to the brevetoxin exposure was evident for children, senior citizens and asthmatics (Fleming et al. 2005, 2007, Kirkpatrick et al. 2011). Similarly, it could be hypothesized that health disorders occur in especially sensitive people during the last stationary phase of the Ostreopsis bloom. Furthermore, we must consider that the permanently exposed human population acquires some sort of immunization after their first exposure period at the beginning of the bloom. Again, the more sensitive individuals would be expected to have a greater tendency to exhibit symptoms towards the end of the bloom. However, this hypothesis should be taken with caution, given that only one person was affected and this second period of sensitivity occurred during the autumn season when it could be associated with other health problems (e.g. viral and bacterial diseases). As already noted, the lack of specificity of the symptoms associated with the presence of O. cf. ovata has been a major factor limiting the establishment of the link between blooms and the effects on human health (Tichadou et al. 2010). Testing these hypotheses requires further careful medical studies, including prospective epidemiological tests and larger human cohorts stratified according to different levels of exposure to the sea conditions (Hertz-Picciotto 1998).

The superimposition of certain meteorological conditions must be necessary for the health symptoms to occur. In particular, landward winds should favour the transport of aerosolized compounds from the seawater to the beach and the nearby inhabited areas, and be involved in the health impacts (Vila et al. 2008, Kirkpatrick et al. 2010). However, no relevant differences concerning the landward wind component during the two affect periods and the days between them were found (Table 3). Our analysis indicates that transport of aerosols containing irritants was possible during all phases of the bloom. This finding reinforces the hypothesis that the irritating compounds are present in the aerosol within a specific phase of the bloom only.

Another fundamental question concerns the process of aerosolization of the potential irritating agents (cells, fragments of cells or toxins, Casabianca et al. 2013). Physical forcing associated with oceanic processes, such as swell (not studied here) and wind-generated turbulence should be major factors. However, we cannot currently establish a specific wind intensity threshold from the values plotted in Figures 1 and 2, because our data treatment smoothed the absolute number and shows average values only. Water motion could interfere with cell detachment from the benthic substrate, favouring the increase of the planktonic cell population, which in turn may be more easily integrated in the aerosol. In this context, the benthic Ostreopsis population would act as a pool for potential human and environmental effects. In contrast, the planktonic fraction, probably more directly related to the aerosolized agents, could constitute a proxy of the actual health impact risk. In our view, any Ostreopsis-related alert risks should be based on both the benthic and planktonic cell abundances. Until now, a threshold of 30000 cells L^–1 in the water has been used to define the alert risk in some Mediterranean areas (International Conference on Ostreopsis Development, ICOD 2012, round table discussion; Funari et al. 2015) provided that meteorological and hydrodynamic conditions are favourable to aerosol and spray formation. Above this value, prohibition of beach access and bathing has been recommended by authorities in France. This concentration of cells, however, has not always been accompanied by reports of health impacts (Tichadou et al. 2010). Because beach closures could have had associated negative impacts on tourism-related social and economic activities, prohibitions were subsequently replaced by preventive alerts and by increased public information about the risk of exposure. These management strategies were recently published in Italian guidelines (Funari et al. 2015). In our study, the main period of effect in August coincided with the highest Ostreopsis concentrations recorded in the plankton, above 10⁴ cells L^–1, and with constant concentrations of around 10⁵ cells g^–1 FW in the benthos. Thus, our present study supports the alert threshold based on plankton cell abundances and also provides new arguments to include the epiphytic cells densities above our observed level as well (>10⁵ cells g^–1 FW macroalgae). This range is in the order of those discussed at the ICOD 2012 round table discussion. To conclude, however, it is important to understand other unknown factors that might be at play in determining dinoflagellate toxicity. Further investigation is needed to define the factors to be included in any possible alert index. Multidisciplinary research and monitoring are required to prevent or mitigate adverse health effects from exposure to complex aerosolized biotoxins produced by microalgae. In the case of the Karenia brevis, for outbreaks in the coasts of Florida and Gulf of Mexico (Kirkpatrick et al. 2010, 2011Fleming 2011) this approach was addressed over several years and allowed progress to be made towards the development of early warning systems. Information about the presence of a Florida red tide is now available from the Gulf of Mexico HAB Forecast (NOAA Harmful Algal Bloom Operational Forecast System, https://tidesandcurrents.noaa.gov/hab/) in conjunction with a local Beach Condition Reporting System.

Ostreopsis recurrent proliferations at temperate latitudes constitute an emergent environmental and public health threat. The potential impacts on human health also include food-borne diseases due to the ingestion of marine products contaminated with PLTX and analogues obtained in the Ostreopsis bloom–affected areas. The Mediterranean strains of Ostreopsis cf. ovata analysed up to now seem to produce only small quantities of a putative PLTX (Ciminiello et al. 2006) and large quantities of other PLTX-like compounds, namely different ovatoxins (e.g. Ciminiello et al. 2008, 2010, Rossi et al 2010). These PLTX-like compounds have been detected in Mediterranean marine fauna (Aligizaki et al. 2011, Amzil et al. 2012, Biré et al. 2013, Biré et al. 2015), suggesting the potential risk of biotoxin accumulation in seafood. For this reason, the Panel on Contaminants in the Food Chain (CONTAM Panel) of the European Food Safety Authority (EFSA 2009) assessed the risks to human health associated with the presence of PLTX-group biotoxins in shellfish, although the European regulation to monitor PLTX has not yet become established. Another aspect to consider in the many facets of the toxicity attributed to Ostreopsis is the role of certain bacteria such as Vibrio, some of which are known to be responsible for wound infections in people swimming in marine waters. Bellés-Garulera et al. (2016) found a positive correlation between Vibrio and Ostreopsis abundances in Sant Andreu de Llavaneres during a year cycle, and suggested a likely role of vibrios in injury infections suffered by beach users during the bloom. Future research should combine ecophysiological, toxicological and environmental (water and weather conditions) data and characterizations of the Ostreopsis blooms to unveil the important factors and mechanisms of the impact on human health. All this information will be essential to the development of early warning systems.