Influence of Environmental Variables on Gambierdiscus spp. (Dinophyceae) Growth and Distribution

Benthic dinoflagellates in the genus Gambierdiscus produce the ciguatoxin precursors responsible for the occurrence of ciguatera toxicity. The prevalence of ciguatera toxins in fish has been linked to the presence and distribution of toxin-producing species in coral reef ecosystems, which is largely determined by the presence of suitable benthic habitat and environmental conditions favorable for growth. Here using single factor experiments, we examined the effects of salinity, irradiance, and temperature on growth of 17 strains of Gambierdiscus representing eight species/phylotypes (G. belizeanus, G. caribaeus, G. carolinianus, G. carpenteri, G. pacificus, G. silvae, Gambierdiscus sp. type 4–5), most of which were established from either Marakei Island, Republic of Kiribati, or St. Thomas, United States Virgin Island (USVI). Comparable to prior studies, growth rates fell within the range of 0–0.48 divisions day-1. In the salinity and temperature studies, Gambierdiscus responded in a near Gaussian, non-linear manner typical for such studies, with optimal and suboptimal growth occurring in the range of salinities of 25 and 45 and 21.0 and 32.5°C. In the irradiance experiment, no mortality was observed; however, growth rates at 55μmol photons · m-2 · s-1 were lower than those at 110–400μmol photons · m-2 · s-1. At the extremes of the environmental conditions tested, growth rates were highly variable, evidenced by large coefficients of variability. However, significant differences in intraspecific growth rates were typically found only at optimal or near-optimal growth conditions. Polynomial regression analyses showed that maximum growth occurred at salinity and temperature levels of 30.1–38.5 and 23.8–29.2°C, respectively. Gambierdiscus growth patterns varied among species, and within individual species: G. belizeanus, G. caribaeus, G. carpenteri, and G. pacificus generally exhibited a wider range of tolerance to environmental conditions, which may explain their broad geographic distribution. In contrast, G. silvae and Gambierdiscus sp. types 4–5 all displayed a comparatively narrow range of tolerance to temperature, salinity, and irradiance.


Introduction
Ciguatera is a common form of phycotoxin-borne seafood disease caused by the consumption of ciguatoxin-contaminated finfish. It is endemic to the tropical and subtropical Caribbean Sea, and the Pacific, Atlantic, and Indian Oceans. Benthic dinoflagellates in the genus Gambierdiscus produce the ciguatoxin precursors responsible for ciguatera toxicity, and their abundance and seasonality has been closely correlated with ciguatera fish poisoning occurrence [1,2]. Proliferation of Gambierdiscus populations is influenced by many environmental factors; amongst them, regimes of temperature, salinity, and irradiance act as determining factors [3][4][5][6][7]. However, the effects of these parameters on Gambierdiscus growth, particularly with respect to species-specific responses, are not well known across the multiple species in this genus.
Species-specific Gambierdiscus growth data emerged beginning in 2009 [5,7,21,22]. In these studies, inter-specific variability was assessed based on the growth responses of a single strain for each species. Intra-specific variance within each species has yet to be determined, and it is unclear if using multiple strains of Gambierdiscus for each species will yield inter-specific growth response patterns similar to those observed previously.
With respect to salinity, Gambierdiscus typically attains maximum growth in the salinity range of 25-35, and depending on species/strains, growth is possible over a much wider salinity range  under laboratory culture conditions [5,7,23]. Oceanic waters in areas where Gambierdiscus spp. occur are generally restricted to salinities of 34-38; however, Gambierdiscus cells have been reported from areas where salinity levels occasionally are outside of this range, such as near river outlets and enclosed water bodies [24,25].
With regard to irradiance, approximately 10% of full sunlight was previously considered the upper threshold for maximum growth of Gambierdiscus [3,4]. Using irradiance characteristics for multiple species of Gambierdiscus, Kibler et al. [5] suggested~2.5-10% (49-231μmol photons Á m -2 Á s -1 ) of surface irradiance supported maximal growth, with optimal growth extending to 75 m depth in the Caribbean. However, these laboratory results still contrast with field observations of Gambierdiscus in habitats exposed to high irradiances, such as sand flats, drifting algae, and detritus [3,26,27], where irradiance approaches surface sunlight, yet there are dense cell accumulations. Thus far, only one study indicated that Gambierdiscus spp. were notably different in response to irradiance regimes at the species level [5]: G. carolinianus and G. pacificus were least adapted to high irradiance and experienced mortality at~300μmol photons Á m -2 Á s -1 . To better understand how irradiance affects Gambierdiscus growth, data from additional species and strains are needed.
Under experimental culture conditions, Gambierdiscus generally achieves maximum growth at 25-31°C, and cannot survive temperatures below~15-21°C or over~31-34°C [5,7,23,28,29]. Field surveys generally agree with this assessment, with Gambierdiscus populations and ciguatera incidence primarily reported from environments with a temperature range of 25-30°C [1,28]. However, recent surveys recorded Gambierdiscus cells at extreme temperatures lower and higher than previously reported, e.g, as low as 14°C [30] and~11°C in the temperate Pacific [31], and in the Red Sea [24], where temperatures can reach or exceed 35°C [32]. Clearly, additional studies on temperature optima and tolerances for Gambierdiscus species and strains are needed to help interpret these reports, as well as to enhance our understanding of the distribution and seasonality of species within this genus in the context of climate change.
This study sought to determine the optimal conditions for growth and the tolerances to temperature (16-38°C), salinity (10-60) and irradiance (55-400μmol photons Á m -2 Á s -1 ) of eight Gambierdiscus species/phylotypes: G. belizeanus, G. caribaeus, G. carolinianus, G. carpenteri, G. pacificus, G. silvae, Gambierdiscus sp. type 4 and Gambierdiscus sp. type 5. Multiple strains were examined for five of the eight species tested. Furthermore, this study represents the first examination of the growth responses of G. silvae, and two Pacific ribotypes (Gambierdiscus sp. type 4 and type 5). The growth responses determined from laboratory experiments were compared with the distribution of Gambierdiscus and ciguatera occurrence in an effort to better understand how these environmental variables influence Gambierdiscus growth and distribution.

Ethics statement
The locations of the field studies are not privately owned or protected. No activity during field study involved any endangered species or protected species. Thus no specific permissions were required for all locations/activities for this study.

Source of Gambierdiscus isolates
In the Caribbean, individual Gambierdiscus cells were obtained from the macroalga Dictyota spp. at St. Thomas, USVI (18°20' 7.30'' N, 64°57' 12.24'' W), with the exception of G. carpenteri, which was collected from Halimeda monile at Long Key, Florida Keys (24°49' 36.70'' N, 80°48' 51.53'' W). In the Pacific, cells were obtained from Halimeda spp. and coral rubble at Marakei Island, Republic of Kiribati (2°0' 0'' N, 173°16' 0'' E). Macroalgae sample processing, Gambierdiscus isolation, and culture establishment procedures were similar to Xu et al. [19]. Isolates were maintained in modified K medium in which Tris buffer and silicate were omitted, with incubation at 23°C, salinity of 32, 100μmol photons Á m -2 Á s -1 of light, and 12:12h light: dark photoperiod. Cultures were maintained for 3-12 months to acclimate to laboratory conditions before they were used in the growth experiments. This study included a total of six Gambierdiscus species and two Gambierdiscus ribotypes, the latter of which may represent undescribed species. Details regarding the isolates are listed in Table 1.

In vivo fluorescence and growth rate
Gambierdiscus growth was assessed by in vivo fluorescence using a 10-AU Fluorometer (Turner Designs, USA). Previous studies confirmed a linear correlation between in vivo fluorescence (relative fluorescence units or RFU) and cell densities (biomass), in which increasing fluorescence was associated with increasing cell numbers rather than an increase in fluorescence per cell [3][4][5]7]. Fluorescence measurements were used to plot fluorescence (log) vs. time; the exponential growth phase portion of this curve was then utilized to calculate growth rates following the equation defined in Guillard [33], in which μ (division Á day -1 ) is the growth rate, and N 1 and N 0 represent the fluorescence at times t 1 and t 0 , respectively. In all experiments, a total of four sequential transfers were performed: the first transfer was conducted to allow the culture to acclimate to the environmental conditions, and data from the second-fourth transfers were collected to determine the growth rate.
Culture fluorescence was measured twice every week at 3-to 4-days intervals. To reduce error during the collection of fluorescence measurements, cultures were mixed fully prior to fluorescence reading. This process differed among species; for G. carolinianus and Gambierdiscus sp. type 4, vortexing was required to resuspend the clumped cells; for the remaining cultures, hand-mixing was sufficient to resuspend cells evenly. All species except G. belizeanus were transferred when fluorescence was >50 RFU. In the case of G. belizeanus preliminary results indicated that cells grew poorly when transferred at <70 RFU, and these cultures were thus transferred when cells were at fluorescence >80 RFU.

Salinity experiments
Salinity experiments were carried out in a model I-35 LLVL Percival incubator (Perry, Iowa, USA) under a constant average temperature of 27°C, and the aforementioned irradiance and photoperiod intervals. Salinity levels ranged from 10-60 (10, 15, 20, 25, 30, 35, 40, 45, 50, 55, and 60), which were created by adding Milli-Q water into natural filtered seawater (salinity 32) to reduce salinity, or adding sea salt evaporated from 0.2 μm filtered natural seawater to increase salinity. This salinity-treated medium was then autoclaved for 45 minutes in Teflon bottles. After autoclaving, the salinity was measured again with a hand-held refractometer and enriched to produce modified K medium where the tris component was omitted. The procedures for determining growth were as described for the temperature experiments. Culture tubes in salinity experiments were placed randomly in the incubator to mitigate micro-environmental differences.

Irradiance experiments
Irradiance experiments were performed in an incubator with a constant mean temperature of 27°C. Irradiance levels of 55,110,200, and 400μmol photonsÁm -2 Ás -1 were established using four equidistant shelves in the incubator. Between 0-6 cool white full spectrum fluorescent bulbs were installed in each shelf, and covered by 0-2 layers of nylon window screens to achieve the desired irradiance levels. Irradiance received by the culture tubes was measured using Digital Scalar Irradiance Meter (Model #: QSP-170, Biospherical Instruments Inc., CA, USA) with probe QSL-100 (Serial #: 1275, Biospherical Instruments Inc.). Culture tubes were placed in random locations in the incubator to once again minimize micro environmental disparity.

Temperature experiments
Temperature experiments were conducted in a temperature gradient bar [34] constructed from an aluminum plate (122cm x 23cm x 5cm) with 120 25mm diameter holes evenly spaced in 20 columns, each with six positions for replicates. Heating was provided by a 300 watt cartridge probe inserted in one end of the bar and cooling water was supplied by a circulating water bath (model # 1156D, VWR, PA, USA) through the opposite end to set up a thermal gradient ranging from 16 Lighting from cool white fluorescent bulbs was supplied below the bar, which provided an average irradiance level of 200μmol photons Á m -2 Á s -1 , under a photoperiod cycle of 12h:12h light: dark. The first transfer was used to acclimate cultures to temperature conditions, and growth rates were calculated from three additional sequential transfers over the culture's exponential phase of growth.

Data analysis
To assess growth variability and statistical differences in growth rates within each Gambierdiscus species in all experiments, coefficient of variation (CV) among conspecific strains was calculated; depending on normality and homoscedasticity of growth rate data, a one-way analysis of variance (ANOVA) (normality, homoscedasticity), Welch's ANOVA (normality, heteroscedasticity), Mann-Whitney Test (2 Independent Samples, non-normality and homoscedasticity) and Kruskal-Wallis H Test (K!3 Independent Samples, non-normality and homoscedasticity) were performed using SPSS statistics 21 (SPSS Inc., IL, USA). In addition, to better understand the growth potential of each Gambierdiscus species/strain, growth rate and salinity, and growth rate and temperature were polynomial fitted using OriginPro 9.0.0 (OriginLab Corporation, MA, USA). The polynomial equations were further analyzed using Matlab R2013a 8.1.0.604 (Mathworks Inc., MA, USA) to calculate the characteristic points where growth maxima, optima, and limitation occurred. The growth characteristics were described by the following parameters similarly defined in Kibler et al. [5]: μ m , maximum growth rate; T m /S m , temperature/salinity of maximum growth; T opt /S opt , temperature/salinity of optimum growth range (μ ! 0.8 × μ m ); T o /S o , the lower temperature/salinity limit for growth, and T u /S u , the upper temperature/salinity limit for growth. Due to data limitations, polynomial regressive analysis was not performed in the irradiance experiment.

Salinity
Growth rate. Growth responses of Gambierdiscus spp. to different salinity at a constant mean temperature of 27°C are shown in Figs 1 and 2. Gambierdiscus growth curves showed a general increase in growth rate with increasing salinity values, followed by a decline at high salinities (Fig 1). Gambierdiscus cells generally did not grow in salinities of 20 or !50; however, certain strains of G. caribaeus and G. carpenteri survived at a salinities of~15 and >50, and strains of G. belizeanus (BP Mar10_6), G. caribaeus (SH Nov09_3), and G. carolinianus (SH Mar10_12) exhibited low growth at these salinity levels (Fig 1).
Growth rates during the salinity studies ranged from no growth to 0.48 division day -1 ; maximum growth rates for each clone were observed at salinities of 30-40, but differed among species as shown in Fig 1. Gambierdiscus carolinianus and G. pacificus exhibited the highest average maximum growth rates, whereas Gambierdiscus sp. type 5 had the lowest. Gambierdiscus caribaeus and G. carpenteri exhibited broad tolerance to salinity compared with the other species examined in this study (Fig 2).
Polynomial regression analysis. To assess Gambierdiscus growth potential at each salinity, growth rate and salinity were described using the polynomial equation: Y = A + B 1 X + B 2 X 2 + . . . + B n X n . Here, X and Y represent salinity and growth rate, respectively. Most strains were fitted to 3 rd to 5 th order polynomial curves with an R value >0.9 (Fig 3, S1 Table).
Polynomial curves for salinity experiments exhibited a relatively symmetrical bell shape ( Fig  3). In the salinity study, μ m varied from 0.18-0.47 divisions day -1 ; S m fell in the salinity range of 30.1-38.5 ( Table 3). The major difference in salinity response was observed in the optimum salinity range (S opt ), and the ability to maintain growth or survive at extreme salinities (S o and S u ). Generally speaking, strains of G. caribaeus and G. carpenteri had a broad S opt and tolerated extreme salinities, whereas strains of G. belizeanus were tolerant to hypersaline conditions. In contrast, strains of G. carolinianus, G. silvae, and Gambierdiscus sp. types 4-5 were sensitive to extreme salinities.

Irradiance
Growth rate. Net growth was observed for all Gambierdiscus strains across the range of irradiances tested (55-400μmol photons Á m -2 Á s -1 ) under a constant mean temperature of 27°C and a salinity of 32. Growth responses to varying irradiance were similar for all 17 strains; i.e., growth Gambierdiscus spp. Growth and Distribution was slow at 55μmol photons Á m -2 Á s -1 and generally plateaued in the range of 110-400μmol photons Á m -2 Á s -1 . No obvious decline was observed at 400μmol photons Á m -2 Á s -1 (Fig 4).
Of the seven species/phylotypes examined, G. pacificus exhibited highest growth rates at irradiance levels !110μmol photons Á m -2 Á s -1 (Fig 5), and growth of both G. pacificus and Gambierdiscus sp. type 5 increased sharply when irradiance increased from 55 to 110μmol photons Á m -2 Á s -1 (Fig 5), suggesting low tolerance of low irradiances. In contrast, the slope of growth rate responses of G. caribaeus and G. carolinianus was comparatively flat (Fig 5), reflecting survival and growth at lower irradiance levels.
Intraspecific variation. CV values of G. belizeanus, G. caribaeus, G. carolinianus, G. pacificus and Gambierdiscus sp. type 4, varied from 0.5-89.9% (Table 4), and high CV values were typically seen at the lowest/highest irradiance levels (Table 4). Most intraspecific variability observed in the irradiance data was not statistically significant (α>0.05). However, for G. belizeanus, G. caribaeus, and G. pacificus, intraspecific growth rates were significantly different at the extreme irradiance levels of 55 or 400μmol photons Á m -2 Á s -1 (Table 4).  Table 2. Gambierdiscus strain numbers, growth rate, coefficient of variation among strains, and intraspecific differences for each species and salinity    Temperature Growth rate. As indicated in Figs 6 and 7, Gambierdiscus species responded to increasing temperatures in a Gaussian mode with enhanced growth to a threshold where rates were maximum, followed by plateau and then decrease in growth.
Gambierdiscus cells generally did not survive at temperatures 17.5°C and ! 32.5°C, although some strains exhibited a narrower range of temperature tolerance (Fig 6). For example, strains 3S0509-27 and 3S0510-19 (G. pacificus) stopped growing at temperatures 18.5°C, and strains of 1D0509-16 and 1D0510-22 (Gambierdiscus sp. type 4) died when temperatures exceeded 31°C (Fig 6). Gambierdiscus spp. appeared to be more sensitive to the higher extreme temperatures than the lower extreme temperatures; growth rates dropped dramatically when temperature approached the upper temperature limit, showing a steep decline in growth (Figs 6 and 7). In particular, G. silvae and Gambierdiscus sp. type 4 exhibited greatest sensitivity to high temperatures, with growth rates declining markedly between 28-30°C. Vegetative cells were generally present in cultures near the lower temperature limit, and were able to survive for periods of 25 days or more under these conditions, albeit with much reduced growth.
Overall, Gambierdiscus growth rates as a function of temperature varied from no growth to a maximum of 0.44 division day -1 . For each species, the maximum growth rate observed at a given temperature was strain-dependent (Fig 6). Mean growth rates at the species level indicated that G. belizeanus, G. caribaeus, and G. pacificus had higher growth rates than the other species; furthermore, G. caribaeus and G. carpenteri were most tolerant to lower temperatures, and G. caribaeus, G. carpenteri, and G. pacificus were most tolerant to higher temperatures (Fig 7).
Intraspecific variation. The coefficient of variation (CV) was used to assess growth rate variability within species at each temperature (Table 5). For G. belizeanus, G. caribaeus, G. carolinianus, G. pacificus, and Gambierdiscus sp. type 4, the CV ranged from 3.5-200% across the temperature ranges analyzed (18.5-32.5, 17.5-34.0, 18.5-31.0, 20.0-32.5 and 18.5-30.0°C). Table 3. Gambierdiscus species growth parameters at salinities of 10-60. Individual growth rate measurements were fitted to polynomial curves. The polynomial equations were used for growth parameter estimation: μ m , maximum growth rate; S m , salinity of maximum growth; S opt , salinity of optimum growth range (μ!0.8×μ m ); S o , the lower salinity limit for growth; S u , the upper salinity limit for growth.

Strain
Species  Highest variability was reported at temperatures at the upper and lower limits, but Gambierdiscus spp. tested exhibited significant intraspecific variation as well (α<0.05), mainly at midrange experimental temperatures (Table 5).
Polynomial regression analysis. To evaluate Gambierdiscus growth potential at each temperature, growth rate and temperature were described using the polynomial equation: Y = A + B 1 X + B 2 X 2 + . . . + B n X n (n 5) where X and Y represent temperature and growth rate, respectively. Major strains were fitted to a 4 th to 5 th order polynomial equation with an R value >0.85 (Fig 8, S2 Table).
Gambierdiscus growth parameters of μ m , T m , T opt , T o , and T u differed among species, and within individual species; however, strains exhibiting skewed growth curves generally had higher μ m and T m , narrower T opt range, and lower T o compared with strains with symmetrical growth curve (Table 6). Among the 17 strains tested, μ m ranged from 0.14 to 0.41 division day -1 , with 0.14 division day -1 measured in BP May10_1 (G. carolinianus) showing a symmetrical μ-°C shape, and 0.41 division day -1 in BP Mar10_7 (G. belizeanus) exhibiting a skewed μ-°C shape.
Gambierdiscus carpenteri required the highest T m to realize maximum growth, in comparison with G. silvae, which required the lowest T m ( Table 6). In addition, species of G. belizeanus, G. caribaeus, and G. pacificus generally exhibited a high μ m , versus Gambierdiscus type 5, which exhibited a low μ m (Table 6). Furthermore, species of G. caribaeus, G. carpenteri, and G. pacificus typically had a wide range of T opt ; G. caribaeus, and G. carpenteri displayed a low T o ; and G. belizeanus, G. caribaeus, G. carpenteri, and G. pacificus displayed a high T u (Table 6). These species thus tolerated extreme temperatures better than others. In contrast, species of Table 4. Gambierdiscus strain numbers, growth rate, coefficient of variation among strains, and intraspecific difference for each species and irradiance (55-400μmol Á photons Á m -2 Á s -1 ).

Discussion
This study examined the growth response patterns of multiple strains of eight Gambierdiscus species/phylotypes under different salinity, irradiance, and temperature, including the first characterization of G. silvae and Gambierdiscus sp. type 4-5. All strains used were isolated from geographically distinct areas compared with strains used in previous studies. The results showed that environmental variability in salinity, irradiance, and temperature can greatly influence Gambierdiscus growth, which was revealed by both intraspecific and interspecific variation. In general, strains of G. belizeanus, G. caribaeus, G. carpenteri, and G. pacificus exhibited a wider range of tolerance to extreme environmental conditions than the other species, consistent with their broad geographic distribution. The growth response of  Gambierdiscus to environmental parameters is clearly a major determinant of the species' abundance and distribution in natural ecosystems, and is useful in evaluating and understanding current and future species distributions and population dynamics both within systems and across geography (e.g. [5]).

Growth rates
Compared with planktonic dinoflagellates, the epibenthic genus Gambierdiscus is slow-growing. Maximum growth rates are generally lower than 0.5 division day -1 [35] and growth rates of approximately 0.3 division day -1 are commonly observed [36], consistent with this study. Thus far, the highest Gambierdiscus growth rate reported is 0.79 division day -1 (0.55/day) for a Hawaiian strain [37], and under optimum combinations of temperature, salinity and light, a growth rate of >0.5 division day -1 was also possible for a Florida strain, GT600 [3]. Unfortunately, the species used in these studies are unknown. Growth responses of G. belizeanus, G. caribaeus, G. carolinianus, and G. pacificus analyzed during this study were similar to previous reports; however, G. carpenteri exhibited a narrower range of growth than in previous studies (see comparison in Table 7). Most of these reported growth rates were measured over a range of optimal growth regimes of salinity, irradiance, and temperature (Table 7). Kibler et al. [29] measured growth at 15-34°C, salinity of 33, and 50-100μmol photonsÁm -2 Ás -1 , while Yoshimatsu et al. [7] measured at 15-35°C, salinity of 20-40, and 90-100μmol photonsÁm -2 Ás -1 . Disparities in growth rates in these experiments most likely reflect genetic differences in growth response rather than culture conditions. Additionally, growth rates of Gambierdiscus determined in this study varied among species, and within individual species (Table 7), even for strains that were isolated from the same location in the same survey (e.g., 1D0509-16 and 1D0510-22). These findings contrast with observations by Bomber et al. [26], who found no significant difference in growth rates among strains from the same station, and concluded that one strain per site was probably representative. Our findings are more similar to results reported by Boyd et al. [38] and Burkholder [39], who demonstrated that it is misleading to use a single strain to represent a phytoplankton functional group.
The aforementioned Gambierdiscus growth pattens add complexity to ciguatera prediction and its management, particularly with respect to model development. Currently it is unknown whether Gambierdiscus toxin production is strain-dependent, but it certainly seems likely. If Gambierdiscus toxin production is stable within species, ciguatera monitoring efforts may be best focused on toxic species with high growth rates. However, if Gambierdiscus toxicity is strain-dependent, as we expect, developing an effective ciguatera monitoring and prediction program will be more difficult. Further efforts are currently underway to determine the species and strain variability of toxin production in Gambierdiscus. Table 6. Gambierdiscus species growth parameters in response to temperature (16-38°C). Individual growth rate measurements were fitted to polynomial curves and equations. The polynomial equations were used for growth parameter estimation: μ m , maximum growth rate; T m , temperature of maximum growth; T opt , temperature of optimum growth range (μ!0.8×μ m ); T o , the lower temperature limit for growth; T u , the upper temperature limit for growth.

Strain
Species Salinity Growth response to salinity. Gambierdiscus growth responses to varying salinity (10-60) were nonlinear with an approximate Gaussian/bell shape (Fig 3). Compared with temperature, the growth curves for salinity were more symmetrical, indicating that Gambierdiscus cells are less sensitive to hypersaline conditions than they are to high temperatures. This is supported by observations by Yoshimatsu et al. [7] that the effect of temperature on growth of Japanese Gambierdiscus was stronger than those of salinity or temperature-salinity varying together.
Intraspecific variability in the salinity experiments was similar to those observed in temperature experiments, i.e., Gambierdiscus cells exhibited larger CV values near either end of the salinity range than in the middle. Growth potential appears to be significantly different within species, especially in the vicinity of optimum salinity (Table 2), which provides further evidence of intra-specific physiological diversity. According to Boyd et al. [38], the distribution and expansion of an organism in neritic waters largely depends upon intraspecific variability in response to temperature. Similarly, the intraspecific variability in growth responses to salinity may help explain why Gambierdiscus spp. are widely distributed in the tropical, subtropical, and temperate regions. In addition to temperature tolerance, the growth response of species of G. belizeanus, G. caribaeus, and G. carpenteri to different salinities under laboratory conditions provides physiological evidence as to why these species are widely distributed [20].
Salinity and Gambierdiscus abundance. Gambierdiscus generally prefers high, stable salinities of 28-35 (summarized by [6]), though this estimate was recently updated to include a broader range than was previously reported [5,7]. Salinities of 34-38 are typical for oceanic waters in areas with ciguatera, thus oceanic salinity should sustain maximum growth of most Gambierdiscus species. This is supported by field observations; for example, no relationship between Gambierdiscus abundance and water salinity was found in French Polynesia, where salinities ranged from 34.3-36.1 [1], which was optimal for Gambierdiscus bloom formation. At the Flower Garden Banks National Marine Sanctuary in the northern Gulf of Mexico, high biodiversity of Gambierdiscus was observed (i.e., six of the seven Gambierdiscus species endemic to the Caribbean region); again, salinity levels of 34-37 measured during the survey support optimum growth [40].
In contrast with the stable or narrow range of salinity in oceanic regions, some coastal locations such as estuaries and bays are affected by freshwater inputs from precipitation and freshwater discharge from land. At these locations, hyposaline conditions may inhibit Gambierdiscus survival. For example, below a salinity of 14 in the coastal zone in the Gulf of Mexico [41], only G. caribaeus and G. carpenteri have a good chance of survival or growth.
Besides hyposaline conditions mentioned, hypersaline environments pose another challenge for Gambierdiscus growth. These conditions readily arise in restricted water bodies in the tropical and subtropical areas with high evaporation, poor circulation, and low freshwater input such as tropical lagoons, where salinities can easily exceed 40 [42]. In response to hypersaline pressure, only Gambierdiscus isolates from G. belizeanus, G. caribaeus, and G. carpenteri may be able to grow.
In previous ecological surveys, positive or negative correlations between Gambierdiscus and salinity were observed only under extreme salinity conditions (hyposaline and hypersaline). For example, Gambierdiscus was absent from river mouth sites due to low salinity [25,43]. Another widely known observation comes from the Virgin Islands; regional precipitation (lower salinity) was significantly and positively related with Gambierdiscus abundance at inshore stations, with Gambierdiscus population maxima co-occurring with peak rainfall [44]. The apparent paradox between these two reports may be attributed to the different ways precipitation influences Gambierdiscus populations. Freshwater inputs from precipitation could be advantageous due to nutrient inputs if water salinity remains within a suitable range for growth, but could be disadvantageous when hyposalinity conditions occur that are suboptimal for Gambierdiscus growth.
Within the context of climate change, precipitation patterns are predicted to occur in which rainfall is less frequent but more intense, and followed by longer dry periods [45]. This new pattern is expected to favor dinoflagellate growth due to increases in water stratification and the availability of nutrients for growth [45,46]. It is uncertain how Gambierdiscus will respond; as benthic organisms, they are distinct from the phytoplanktonic dinoflagellates in that they are generally associated with a benthic macroalgal habitat. Furthermore, one previous salinity shock experiment revealed that growth responses of Gambierdiscus to instantaneous salinity decreases were species-dependent and included a range of responses such as no effect, slowed growth, or mortality [5]. Further work that includes additional Gambierdiscus species and geographically distinct strains is thus required.

Irradiance
Here, all 17 Gambierdiscus strains tested grew at 55-400μmol photons Á m -2 Á s -1 , and no obvious growth inhibition was observed at 400μmol photons Á m -2 Á s -1 . This pattern contrasts with other laboratory studies examining the irradiance response of several Gambierdiscus species. Clones GT600 and GT600A have been shown to be inhibited at irradiances of >232 and >225μmol photons Á m -2 Á s -1 , respectively (units conversion refers to [47]). Similarly, G. caribaeus growth decreased when irradiance exceeded 300μmol photons Á m -2 Á s -1 , and G. carolinianus and G. pacificus could not survive at 200 and 400μmol photons Á m -2 Á s -1 , respectively [5]. No obvious growth inhibition was observed in this study when Gambierdiscus cells were cultured at 110-400μmol photons Á m -2 Á s -1 . As in the temperature and salinity experiments, intraspecific variability may play a crucial role in explaining the differences among observations. In Kibler et al. [5], all eight Gambierdiscus species exhibited low light adaptation, requiring only 6-17μmol photons Á m -2 Á s -1 to maintain growth. Here, under the lowest irradiance tested, 55μmol photons Á m -2 Á s -1 , all eight Gambierdiscus species/phylotypes examined grew and no mortality was recorded; light tolerance of <55μmol photons Á m -2 Á s -1 is thus likely.
There is an apparent inconsistency between laboratory findings and field observations of Gambierdiscus response to light intensity. The genus Gambierdiscus typically attains optimum growth at~10% of full sunlight [4] or maxima growth at~2.5-10% of surface irradiance [5], which agrees well with the general irradiance requirement for dinoflagellates [48]. Interestingly, growth responses to irradiance observed in this study were diverse, which may help explain why data obtained experimentally does not always reflect field observations of Gambierdiscus in shallow environments subjected to high light intensities. This study determined that the optimum light intensity for growth was~4.4-16% of full light (full sunlight = 2500μmol photons Á m -2 Á s -1 ), which is higher than reported previously. However, it does not follow ecological observations in which Gambierdiscus cells were detected in the shallow waters of 1-5 m [49], on sparse macroalgae and bright sand flats [3], on drifting seaweed [10,26], or floating detritus [27]. The tolerance for high-irradiance environments may be partly attributed to the finely-branched and three-dimensional structure of host macroalgae, which provides substrate for Gambierdiscus attachment and shields cells from strong light damage [50]. This hypothesis was indirectly verified by PAM fluorescence in that Gambierdiscus spp. exhibited typical characteristics of "shade-adapted" organisms [51].
Another explanation is that like other benthic dinoflagellates such as in the genus Ostreopsis [52], Gambierdiscus cells tend to produce more mucus at higher irradiance levels. This mucus production causes cell aggregation and a cell complex enveloped by large quantities of mucus protects cells from high light due to self-shading. Besides the physical structures that Gambierdiscus spp. utilize to shade themselves from high irradiance, photoprotection mechanisms observed in other dinoflagellates such as secretion of UV radiation-absorbing compounds and changes in pigment composition should also be considered [53,54].
Irradiance plays an important role in defining Gambierdiscus vertical distribution in the water columns [5]. In most prior field studies, the genus Gambierdiscus was largely collected from shallow water depths <5 m [49], but were also recorded at depths of 10-40 m [40,55]. Thus far, the maximum depth reported was for G. carolinianus, which was collected from 45.7 m in the northern Gulf of Mexico [40]. These findings are supported by other studies indicating that irradiance levels of 6-17μmol photons Á m -2 Á s -1 are sufficient to maintain Gambierdiscus growth, corresponding to >150m depth in tropical waters [5]. Since both temperature and irradiance decrease with increasing water depth, Gambierdiscus spp. in deep waters are not likely to experience conditions for optimal growth; however, Gambierdiscus cells inhabiting these ecosystems may serve as source populations for surrounding shallower niches.

Temperature
Growth response to temperature. Although a linear relationship has been observed between Gambierdiscus growth responses and some environmental parameters such as DIN and phosphate [23], the growth responses to temperature we observed were near Gaussian in shape, with some strains exhibiting a more symmetrical shape than others (Fig 8). This near Gaussian response, seen frequently in similar studies of other phytoplankton species, indicates that Gambierdiscus growth is optimal and suboptimal within discrete temperature ranges, which is consistent with many previous observations [3][4][5]29]. For G. caribaeus, the symmetric or slightly skewed growth response curve observed in this study is similar to that in Tester et al. [28] and Kibler et al. [29]. However, this response is markedly different from G. caribaeus in Kibler et al. [5], which had a highly skewed growth rate shape in response to temperature. For G. belizeanus, G. carolinianus, and G. pacificus, growth response shapes reported here and those described by Kibler et al. [5,29] are not identical but can be regarded as analogous.
Our findings, together with others [5,23,28,29], suggest that the global distribution of G. belizeanus, G. caribaeus, and G. carpenteri may be due in part to their broad tolerance to environmental conditions, especially to temperature. It is not surprising that G. pacificus also exhibited a wide thermal tolerance, as that species is frequently observed and is broadly disturbed in the tropical Pacific. Regarding G. carolinianus, our data agree with previous work indicating that this species has a relatively low maximum temperature for growth (T m =~25-27.1°C) and is well-adapted to lower temperatures (T o = 15.8°C) [28,29]. This species has been isolated as far north as North Carolina on the USA east coast, near 34°N [9] and collected from the deepest site ever recorded (45.7 m) for the genus [40]. The temperature parameters T m and T u of G. carolinianus are similar to previous results: 23.8-27.0°C vs.~25-27.1°C and 30.3-31.6°C vs. 32.5°C, respectively [28,29]; however, T o in the current study was much higher (18.4-18.6°C vs. 15.8°C) [29]. These differences suggest that G. carolinianus originating from St Thomas, USVI may be more sensitive to lower temperatures than the strain isolated from North Carolina. It also implies that differences among strains within each species may reflect geographic origin, and cautions against using one strain to represent the physiological characteristics of a species.
To our knowledge, this study represents the first characterization of the relationship between temperature and growth for G. silvae and Gambierdiscus types 4 and 5. To date, only Gambierdiscus sp. ribotype 2 and types 2-3 were used in experiments assessing growth response to temperature. Similar to the findings of this study, Gambierdiscus sp. ribotype 2 and type 3 had a narrow T opt , corresponding to 24.5-30.1 and 22-25°C, respectively [7,29]. But Gambierdiscus sp. type 2 showed a T opt of 21-28°C [7]. Subsequent work examining additional strains will determine if similar patterns of intra-specific diversity also exists within these ribotypes.
The majority of growth studies carried out on Gambierdiscus used a single strain to represent each species [5,7,28]; however, like Kibler et al. [29], we examined multiple strains per species. In the temperature experiments, we observed intraspecific variability, including the response shape, and all parameters -μ m , T m , T opt , T o , and T u . Although the intraspecific CV of growth rate varied most at either end of the temperature range, statistically significant intraspecific variation was encountered within the T opt range. This agrees with Tindall and Morton [49], who noted that competitive growth rates of Gambierdiscus and other ciguatera associated dinoflagellates could only be compared near the temperature for optimal growth. Our results are also similar to those documented by Boyd et al. [38], who reported that CV values among strains in both diatoms and dinoflagellates are lowest near the optimal growth temperature, and highest at the extremes of temperature tolerance, suggesting strong genotypic selection pressure at these end points [38].
Temperature and Gambierdiscus distribution. Temperature plays an essential role in restricting the distribution of epiphytic dinoflagellates. The genus Gambierdiscus generally has an optimal temperature range of 19-31°C, and sustains growth between 15-34°C (this study, [5,7,29]). The optimum temperature for oceanic phytoplankton is considered to be strongly related to the mean environmental temperature that species are exposed to [56]; the high temperature requirement for Gambierdiscus growth thus explains why this organism and ciguatera incidence are circumtropically endemic.
The parameter T o indicates the extent to which Gambierdiscus is adapted to low temperatures and helps to define its latitudinal distribution. The growth measurements of 17 Gambierdiscus strains in this study produced T o estimates varying from 15.9-20.1°C, which are close to previous laboratory reports, such as [16][17][18][19].5°C for Florida Keys isolate GT600 [3] and 15.0-20.0°C for multiple species/strains [5,29]. Field surveys and laboratory culturing, however, show that Gambierdiscus species can survive in a broad range of temperate environments. Along the Pacific coast in Japan, Gambierdiscus cells occurred year-round in locations where the wintertime temperatures drop to as low as~11°C [31], which represents the lowest field temperature recorded thus far for the genus. Under laboratory conditions, Gambierdiscus cultures established from the temperate coasts in Japan survived for at least three months at 10°C [31].
Using the lowest T o , 15.9°C from BP Aug08 (G. caribaeus), and wintertime Sea Surface Temperature (SST) (http://www.nodc.noaa.gov/about/oceanclimate.html) to predict its possible distribution boundary, this G. caribaeus strain may extend as far north as~38.0°N, and as far south as~45.0°S. This predicted range is narrower than the one described in Kibler et al. [5], who concluded that the greatest latitudes of distribution are approximately 38.5°N, and 47.1°S. The difference between these two predictions of Gambierdiscus distribution is attributed to the different T o values used; in Kibler et al. [5], T o was 15°C (NOAA 6, G. carolinianus) versus 15.9°C from our study. Clearly, a 1.0°C difference in T o results in a marked change in Gambierdiscus' predicted latitudinal range. However, the boundaries predicted in this study or that of Kibler et al. [5] are still broader than the observed field distribution of Gambierdiscus; i.e. 35°N-37°S [6]. This suggests that continued field sampling may discover additional Gambierdiscus populations beyond this range.
The temperature parameter T u is another important index for determining how Gambierdiscus responds to high temperature. Estimates of T u fell within 29.8-35.4°C, a broad range that encompasses previous T u records, such as 31.1-35.1°C and 25-30°C for multiple isolates in Kibler et al. [5,29] and Yoshimatsu et al. [7], respectively. The summer oceanic SST is generally within the T u range of Gambierdiscus (http://www.nodc.noaa.gov/about/oceanclimate. html), with the warmest areas primarily located in the tropical Caribbean Sea, and the Pacific and Indian Oceans. Within these regions, the Indo-Pacific Warm Pool (IPWP) and West Pacific Warm Pool (WPWP) are two large bodies of seawater whose SST consistently remains above 28.5°C [32]. If the temperatures of the IPWP and WPWP are below T u , Gambierdiscus populations are capable of growth and development, provided that other environmental conditions are suitable for growth. However, once conditions exceed T u for specific Gambierdiscus species or strains, mortality may occur, possibly altering the composition of Gambierdiscus populations in the system. This may help explain why nations in close proximity to the IPWP have low or negligible ciguatera rates even though they are located within the tropics [28,32], generally regarded as ideal habitat for Gambierdiscus. Similar conditions exist in Red Sea and Arabian Sea, where rates of ciguatera are also negligible [32]. Gambierdiscus cells have been observed in these areas [24,57], where seawater temperatures sometimes reach as high as 35°C [32]. On the basis of laboratory results, SSTs at or above 35°C would be lethal to most Gambierdiscus strains.
Gambierdiscus abundance is thought to be closely related to ocean warming. If warming waters still satisfy their growth requirement, a positive correlation between temperature and abundance may be expected [23]. This highlights an important point on semantics; some studies (e.g., [28]) have used observations of increasing ciguatera incidence or increased Gambierdiscus growth rates with increasing temperatures to argue that there will be range extensions or expansions. Clearly, however, if temperatures exceed thermal tolerance thresholds, Gambierdiscus spp. abundance and ciguatera may decline or disappear in some areas [23,32], such that the net effect is not necessarily a range expansion of a species, but rather a shift in that range.
The Intergovernomental Panel on Climate Change (IPCC) estimated that by 2100, ocean SSTs will increase by 0.6-2.0°C, with highest increase occuring in tropical and Northern Hemisphere subtropical regions [58]. Given that the temperature gap between T opt and T u varied from 1.7-3.7°C in this study and 1.2-3.3°C in Kibler et al. [5,29], one may expect that ocean warming in the next 100 years will inhibit those Gamberdiscus strains with a narrow T opt -T u gap (e.g., G. carolinianus and G. pacificus, Table 6). Additionally, strains with a relatively low T u (e.g. BP Mar10_6 and SH Mar10_12, Table 6) may be inhibited by this warming. Ocean warming is expected to result in declining tropical phytoplankton diversity, as many tropical strains, in the absence of evolution, are unable to survive even small increases in temperature [56]. The epibenthic dinoflagellate Gambierdiscus is no exception due to its sensitivity to the upper temperature range (T u ), and narrow gaps between T opt and T u . However, seawater does not have an infinite capacity to warm, and a phenomena called the tropical thermostat may suppress ocean warming [59]. As a result, the future relationship between Gambierdiscus and ciguatera due to rising seawater temperatures is complex. Nevertheless, under conditions of warming, Gambierdiscus populations will likely proliferate in some areas, and possibly decrease in regions where ocean temperatures exceed T u .
Supporting Information S1 Table. Growth rate statistics calculated with polynomial fit of Gambierdiscus growth rates in response to salinity (10-60). (DOC) S2 Table. Growth rate statistics calculated with polynomial fit of Gambierdiscus growth rates in response to temperature (16-38°C). (DOC)