Population fluctuations of Pyrodinium bahamense and Ceratium furca ( Dinophyceae ) in Laguna Grande , Puerto Rico , and environmental variables associated during a three-year period

Fluctuaciones poblacionales de Pyrodinium bahamense y Ceratium furca (Dinophyceae) en Laguna Grande, Puerto Rico, y variables ambientales asociadas durante un periodo de tres años. Bioluminescent bays and lagoons are unique natural environments and popular tourist attractions. However, the bioluminescence in many of these water bodies has declined, principally due to anthropogenic activities. In the Caribbean, the bioluminescence in these bays and lagoons is mostly produced by the dinoflagellate Pyrodinium bahamense var. bahamense. Laguna Grande is one of the three year-round bioluminescent water bodies in Puerto Rico that are known to remain but P. bahamense var. bahamense density fluctuations have not been studied. In this study we describe water quality parameters and density fluctuations of the most common dinoflagellates in Laguna Grande, P. bahamense var. bahamense and Ceratium furca, over a three-year period. For this, three sampling stations were established in Laguna Grande from which water samples were collected in triplicate and analyzed for temperature, phosphates, nitrates, salinity, water transparency, fluorescence, and dinoflagellate densities, at the water surface and at 2m depth, from May 2003 to May 2006. The results showed a density fluctuation pattern for P. bahamense var. bahamense, where higher densities were observed mainly from April to September, and lower densities from October to February. Density fluctuations of C. furca were more erratic and a repetitive pattern was not observed. Densities of P. bahamense var. bahamense ranged from 0.48 to 90 978 cells/L and densities of C. furca ranged from 0 to 11 200 cells/L. The mean population density throughout the sampling period was significantly higher in P. bahamense var. bahamense (mean=18 958.5 cells/L) than in C. furca (mean=2 601.9 cells/L). Population densities of P. bahamense var. bahamense were negatively correlated with C. furca densities during the first year of sampling; however, they were positively correlated during the third year. Non-significant differences between surface and 2m depth samples were observed for temperature, phosphates, nitrates, salinity, fluorescence, and densities of P. bahamense var. bahamense and C. furca, suggesting a vertically mixed water column. Water transparency was positively correlated with salinity and negatively correlated with fluorescence. Fluorescence was negatively correlated with salinity. The mean population densities of P. bahamense var. bahamense and C. furca observed in this study were within the range of previous reports in other bioluminescent water bodies in Puerto Rico and Florida, USA. In order to conserve the continuous P. bahamense var. bahamense populations in Laguna Grande, as well as its bioluminescence, it is recommended to maintain the existing water flow levels in the 1.5km long inlet/outlet channel; to maintain unpolluted water quality parameters within the bay, the hydrographical basin and adjacent waters, and to preserve mangrove communities within the basin and adjacent areas. Results of this study may help to develop management plans aiming to conserve P. bahamense, its bioluminescence and the lagoon attraction. Rev. Biol. Trop. 61 (4): 1799-1813. Epub 2013 December 01.

Bioluminescent bays and lagoons are economically important as tourist attractions (González-Sánchez, Muñoz-Salinas & Roset, 2012).The bioluminescence in many of these water bodies have declined or failed due to anthropogenic activities (e.g., Fire Lake, New Providence Island, Bahamas; Oyster Bay, Jamaica ;Environmental Solutions, Ltd., 2005).There is some disagreement about the number of year-round bioluminescent bays in the world, but at least three of those remaining functional are found in Puerto Rico.Bioluminescent water bodies appear to be threatened due to poor water quality, habitat deterioration, light pollution (causes poor perception of bioluminescence) and unwise management.Conservation of these systems is important because they are rare and unique.
Dinoflagellate bioluminescence is a defensive behavior that reduces grazing (Esaias & Curl, 1972;White, 1979) by affecting the behavior of animals that feed on them (Buskey & Swift, 1983;Buskey et al., 1983).In addition, it may serve as a "burglar alarm" to attract a secondary predator that threatens to eat the primary predator (Morin, 1983;Young, 1983).Margalef (1961) proposed a general water circulation mechanism for Bahía Fosforescente (Puerto Rico), in order to explain the retention of P. bahamense var.bahamense, as well as other phytoplankton.Inshore winds, aimed towards the inside of bioluminescent bays, can also create hydrological conditions leading to high water retention times in the shallower innermost portions of these water bodies (Seliger et al., 1971).This could also contribute to the retention of nutrients and phytoplankton.
Even though the presence of P. bahamense var.bahamense has been reported for Laguna Grande (Candelas, Cintrón & McKenzie, 1968;Zayas, 1979), a highly regarded (Departamento de Recursos Naturales de Puerto Rico, 1984) and one of the most frequently visited bioluminescent lagoon in Puerto Rico, no study has been done to describe the population fluctuations of P. bahamense at this site.
The goal of this study was to describe water quality parameters on density fluctuations of P. bahamense var.bahamense and C. furca, the two most abundant dinoflagellate species in Puerto Rico bioluminescent bays and lagoons, over a three year period, at surface and 2m depth sites in Laguna Grande, Puerto Rico.Concurrently, to gather basic environmental and ecological information in order to understand better the unique ecosystem.This will help to make more realistic and sound management plans in order to conserve the bioluminescence and water quality in the lagoon.

MATERIALS AND METHODS
Study site: Laguna Grande is a coastal lagoon located on the North-East coast of Puerto Rico.It is situated within a basin-type mangrove forest (sensu Lugo & Snedaker, 1974) and bordered by Rhizophora mangle (Linnaeus, 1753).The Lagoon is surrounded by the Cabezas de San Juan Natural Reserve, managed by the Conservation Trust of Puerto Rico (Fig. 1).
The average depth of Laguna Grande is 3m and its maximum depth is 5m.It occupies an area of about 50ha and contains about 662 000m 3 of water (Soler-López & Santos, 2010).The South-Eastern portion of Laguna Grande is connected to Las Croabas Bay by the Laguna Grande Channel, a 1.5km long channel (or tidal inlet) that averages about 5m across  and 1m deep (Zayas, 1979;Weaver, Ramírez & Coll, 1998;Soler-López & Santos, 2010).Zayas (1979) reported salinity concentrations ranging from 20.5 to 36.2psu (mean=33.6psu),oxygen concentrations ranging from 2.0 to 7.2mg/L (mean=5.3mg/L),and phosphate concentrations ranging from 0.01 to 0.10μg-at /L (mean=0.03μg-at/L), for stations located in the center of the lagoon.The average flushing rate of Laguna Grande is 7.7d (Soler-López & Santos, 2010).
Field methods: Three sampling stations were established in the central portion of Laguna Grande, using a Garmin® GPS model 73 (Station 1: 18°22'31.69"N -65°37'28.98"W; Station 2: 18°22'34.93"N -65°37'22.24"W; Station 3: 18°22'40.32"N -65°37'26.31"W; Fig. 1).At each station, water samples were collected in triplicate using 500mL bottles, at the surface and at 2m depth (the maximum depth at mean low tide in one of the stations is approximately 2.4m).Samples were collected once or twice every month during a period of approximately three years, from May 3, 2003to May 25, 2006.The Surface samples were collected at approximately 10cm below the surface by grab sampling.A manual diaphragm water pump connected to a 2.5cm diameter flexible tube, attached to a 3.5m-long pole, was used to collect the 2m depth water samples.The water pump was flushed with water from each station before collecting each sample.After sample collection, bottles were covered with aluminum paper and immediately placed under ice in a cooler.
The water temperature in each sample was determined immediately after collection using an alcohol laboratory thermometer.Water transparency was determined at each station using a 20cm diameter black/white quadrant limnological Secchi disk.All samples were collected during the daytime.The field sampling procedures lasted approximately two hours.
Laboratory methods: Water samples were taken to the laboratory within one hour after performing the field sampling procedures.
Sample bottles were mixed by inversion in order to distribute the sample homogeneously.One 250mL sub-sample was taken from each sampling bottle for each physico-chemical parameter to be analyzed.
Chlorophyll-a was analyzed using a Turner Designs Model TD-700 fluorometer, using the In-Vivo method (Turner Designs, 1999).The fluorometer was configured with a 340-500nm excitation filter, a 680nm emission filter and a Blue Mercury Vapor lamp.Borosilicate Test Tubes (13x100mm) were used to perform the analyses (Turner Designs, 1999).All blanks and samples were read at 25°C.Relative fluorescence measurements were reported as Relative Fluorescence Units (RFU).The fluorometer was calibrated prior to each use by means of a blank and a solid secondary standard (Turner Designs, 1999).
Salinity was determined using a Fisher model 13-946-27 refractometer, calibrated with distilled water prior to analysis.Daily rainfall data was obtained from NOAA's Paraíso station, located approximately 13km Southwest of the study area.Rainfall data was recorded using a Fischer Porter Rain Gauge.Cumulative precipitation was calculated for 3, 6, 9, 12, 15 and 18 day intervals before each sampling date.
In order to preserve samples, each sampling bottle was mixed several times by gentle inversion.Immediately afterwards, a 248.5mL sub-sample was poured into a modified sedimentation chamber and preserved in 1% Lugol's solution, making a final volume of 250mL.The preserved subsamples were allowed to settle by gravity for three days (Wetzel & Likens, 2000).Each subsample was concentrated to 100mL by carefully removing the supernatant with a pipette.The supernatant was gravity filtered through a 20µ Nitex® Nylon Bolt Cloth, and the cloth inspected for the presence of phytoplankton.If organisms were observed in the cloth, these were returned to the sedimentation chambers.
After mixing each 100mL concentrated plankton sample, one mL subsample was obtained, with an automatic volumetric pipet, and transferred to a Sedwick-Rafter (S-R) counting cell (Wetzel & Likens, 2000).The S-R cell was covered with a coverglass and all P. bahamense var.bahamense and C. furca were counted under a Nikon Eclipse E-600, or a Leica CME microscope, at 100 or 200x magnification.Each subsample was counted once.
A main effects three-way ANOVA was used to evaluate differences between months, sites and depths, for population densities of P. bahamense var.bahamense and C. furca, and for all physical-chemical parameters; except Secchi depth, which was evaluated using a two-way ANOVA (Secchi depth was measured vertically in the water column, therefore depths effects are not applicable).Data were transformed with log e, log 10 and square root to try to conform to assumptions of ANOVA.Levene's tests were used to assess homoscedasticity in transformed and non-transformed data.All ANOVA tests were carried out only using homoscedastic data.Kruskal-Wallis nonparametric tests were applied to variables that showed significant heteroscedasticity, even after been transformed.We assume independence among observations in the dataset.
A multiple correlation was used to determine possible relationships among all variables (Sokal & Rohlf, 1994).Statistical analyses were performed using IBM® SPSS® Statistics 19 software (http://www.spss.com).Any correlation equal to, or less than 0.5 was considered significant.
Salinity: Salinity ranged from 18 to 42 practical salinity units (psu, Fig. 2).During strong rainfall events decreases in salinity were observed both in surface and 2m depth waters.During less severe events, decreases in salinity were observed only in surface waters.ANOVA detected significant differences between months (p<0.001) and depths (p=0.001) but not between sites (p>0.05,Table 2).
Rainfall: Daily rainfall at Paraíso Station ranged from 0 to 10.4cm (Table 1).Cumulative 9-d, 12-d, 15-d and 18-d rainfall data (cumulative amount of rain recorded prior to each sampling occasion) was negatively correlated with salinity.Also, cumulative 6-d, 9-d, 12-d, 15-d and 18-d rainfall data was negatively correlated with fluorescence.Cumulative 9-d rainfall was negatively correlated with Secchi depth; and cumulative 3-d rainfall was correlated with phosphates (Table 3).
The lowest salinity concentrations (18psu, Table 1), recorded during September 21, 2004, corresponded to the passage of Hurricane Jeanne on September 14 (22.9cm of rain in 24hr at the Paraíso station).The second lowest salinities, recorded during November 21, 2003, were caused by the heavy rains received during the occurrence of a tropical wave in November  2).Even though periods of 7 and 8d had passed between the passage of Hurricane Jeanne, the tropical wave; and the sampling dates after those events, low salinity readings were still recorded at Laguna Grande.
Fluorescence: The highest peaks in fluorescence (Table 1) were observed during the low salinity dates of November 21, 2003;September 21, 2004;and May 26, 2005 (Fig. 4).During the first two peaks fluorescence was higher in the surface; however, during the third peak, higher readings were observed in the 2m depth samples.Fluorescence was negatively correlated with salinity (r=-0.802,df=40, p<.0001).ANOVA detected significant differences between months (p<0.001) but not between depths or sites (p>0.05,Table 2).Densities of P. bahamense var.bahamense and C. furca: ANOVA detected significant differences between months but not between stations or depths, for densities of P. bahamense var.bahamense and C. furca (Table 2).
Densities of P. bahamense var.bahamense ranged from 0.48 to 90 978 cells/L (Figs. 5 and   6).Even though no seasonality was observed for P. bahamense var.bahamense density fluctuations, higher densities were observed from April to September and lower densities from October to February.

DISCUSSION
Laguna Grande has less water interchange with the ocean than the bioluminescent bays Puerto Mosquito, Bahía Fosforescente and Bahía Monsio José; because of the presence of the shallow (average depth approximately 1m) and narrow 1.5km long tidal inlet/outlet channel, that limits water interchange between the lagoon and the ocean (Fig. 1).The restricted water flow into the lagoon causes a 3.5hr delay between the tidal peaks in the ocean and the corresponding tidal peaks inside the lagoon (Soler-López & Santos, 2010).It is visibly evident that the currents in the channel can cause strong turbulence and mixing in the water column.Therefore, the "negative" estuary (or hypersaline estuary) type of circulation, described by Margalef (1961) and further discussed by Seliger et al. (1971), does not occur in Laguna Grande.The average residence (or flushing) time of Laguna Grande is 7.7d (Soler-López & Santos, 2010).Most other bioluminescent lagoons or bays in Puerto Rico should have shorter water residence times because of their wider and less restrictive entrances.Therefore, the retention of nutrients, humic substances and microalgae should be longer in Laguna Grande.Also, a broader range in physical-chemical parameters is expected.Bays and/or lagoon habitats where P. bahamense var.bahamense dominate are characterized by relatively long residence times, both in Florida (Phlips et al., 2006) and Puerto Rico (Margalef, 1961).
Since the geomorphological and hydrological characteristics of Laguna Grande can lead to water accumulation in the lagoon, we calculated cumulative rainfall (3,6,9,12,15 and 18 day intervals before each sampling date, using NOAA Paraíso station data) and correlated it with the physical-chemical parameters observed in this study.The resulting correlation coefficients measured the degree of association between cumulative rainfall and each of the other parameters.The various significant correlations between cumulative rainfall; salinity and fluorescence could be due, in part, because of the relatively low water interchange between Laguna Grande and the ocean.The effects of cumulative rainfall on salinity and fluorescence could be significantly detected for up to 18d.The strong negative correlation between salinity and fluorescence can be explained by the washout of algae and leaf particles, from the channels and other areas of the Laguna Grande basin, to the sampling stations located in the central portion of the lagoon.Rainfall events can increase the nitrogen: phosphorus ratio and push the system towards phosphorus limitation (Smalley & Coats, 2002), changing the planktonic species composition.In the Arabian Sea, an increase in the nitrogen: phosphorus ratio, following monsoon rains, favors the diatom Biddulphia sinensis relative to the dinoflagellate C. furca (Qasim, Bhattathiri & Devassy, 1973).In Kuwait's waters (Arabian Sea), lower salinity water, associated with monsoon rains, is correlated with a high prevalence of diatoms whereas higher salinity water is correlated with a higher dominance of dinoflagellates (Polikarpov, Al-Yamani & Saburova, 2009).
The non-significant correlations between the fluorescence values; and the densities of P. bahamense var.bahamense, and C. furca were possibly due to the fact that all the other photosynthetic planktonic microalgae in the lagoon were not considered in these analyses.The densities of P. bahamense var.bahamense and C. furca are only a portion of the total density of all the photosynthetic planktonic microalgae.Therefore, the omission of most microalgae species could explain these non-significant correlations.
The non-significant differences detected in Laguna Grande, between surface and 2m depth samples, for temperature, salinity, phosphates, nitrates, fluorescence, and densities of P. bahamense var.bahamense and C. furca, suggest a vertically mixed water column.Relatively strong trade winds, combined with an average depth of only approximately three meters (Soler-López & Santos, 2010) facilitate the mixing process.However, vertical stratification in salinity and fluorescence, affecting the entire water column, was observed after the most significant rain events, recorded on November 21, 2003 andSeptember 21, 2004.Less severe rain events affected only the salinity in the surface (e.g.May 26, 2005).Soler-López & Santos (2010) also observed some vertical stratification in Laguna Grande.
The correlations between the density of P. bahamense var.bahamense and C. furca were performed to show numerical relationships between these species but not to infer predatorprey or competition relationships.However, significant negative correlations might indicate potentially competing species.The correlation was significantly positive during the first year of our study, significantly negative during the third year, but not significant over the entire three-year period.Similarly, Seixas (1983Seixas ( , 1988)), Walker (1997) and Soler-Figueroa (2006) did not observe significant relationships between these species in other bioluminescent bays/lagoons in Puerto Rico.
Heterotrophic dinoflagellates of the genus Protoperidinium (Hansen & Calado, 1999), observed at low densities in Laguna Grande, might prey on C. furca and/or P. bahamense and may affect their densities.However, we have not detected this interaction in live phytoplankton samples from Laguna Grande.
An attempt was made to compare the range and the mean population densities of P. bahamense var.bahamense and C. furca observed in this study (Table 1), with those reported by Seixas (1983Seixas ( , 1988)), Walker (1997), Soler-Figueroa (2006) and Phlips, Badylak, Youn & Kelley (2004, no data for C. furca).The mean population density of P. bahamense var.bahamense and C. furca in Laguna Grande is within the range of these other studies.The density of P. bahamense var.bahamense is higher than that of C. furca in Laguna Grande and Puerto Mosquito, but mostly lower in Bahía Fosforescente and Bahía Monsio José.The mean population density of P. bahamense var.bahamense in the Indian River Lagoon, Florida (58 934 cells/L; Phlips et al., 2004) is within the range of the mean densities reported in the Puerto Rico sites.
In order to conserve the continuous P. bahamense var.bahamense populations in Laguna Grande, as well as most of its bioluminescence, it is important to maintain the existing water flow levels in the tidal inlet/outlet channel, connecting Laguna Grande to Las Croabas Bay.This will help to keep the existing water exchange rates with the ocean, maintaining the actual nutrient and salinity levels of the lagoon.Even though P. bahamense var.bahamense is considered a euryhaline sub-species, having tolerance salinity limits in Florida of approximately form 10 to 45psu (Phlips et al., 2006), and very likely, similar tolerance limits in Puerto Rico; should the channel close, either by natural or artificial means, the salinity and nutrient levels in Laguna Grande will be altered.This should lead to changes in the planktonic species composition and to the eventual elimination, or population crash, of P. bahamense var.bahamense, as well as many other species non-tolerant to very low or very high salinities.It is also important to maintain unpolluted water quality parameters within the bay, hydrographical basin and adjacent waters; preserve mangrove communities within the basin and adjacent areas; and establish sound management plans.Light pollution should be minimized to increase human perception of bioluminescence.Bioluminescence in the bay is a touristic attraction and the intensity of the biochemical reaction impact the natural experience which influences the local economy.
In order to have more accurate measurements of density variations through time in Laguna Grande it is important to sample at more frequent intervals since the replication rate of P. bahamense var.bahamense and C. furca is much faster than the sampling interval used in this study.

ACKNOWLEDGMENTS
We want to acknowledge the effort given by the following people who have helped during the field sampling, water analyses and/or other aspects of this study: Elizabeth Padilla, Arlyn Fuentes and Sandra Maldonado (Conservation Trust of Puerto Rico); Rufo Vega, Sylvia Vélez, Cedar García, Raymond Tremblay, Juan Morales, Jackie Vega, Jackeline Maldonado, Sofía Burgos, Noraida Martínez, Héctor Ramos, Armando Opio, and Diana Medina (University of Puerto Rico at Humacao); Juan González Lagoa and Brenda Soler Figueroa (University of Puerto Rico at Mayagüez); and three anonymous reviewers.This work was funded, in part, by several grants from Fondo para la Investigación (FoPI), University of Puerto Rico, Humacao.

Fig. 1 .
Fig. 1.Satellite image of Puerto Rico (left) showing the location of Laguna Grande, Puerto Mosquito and Bahía Fosforescente; and a detailed image of Laguna Grande (right), showing the position of the study sites and the inlet/outlet channel.Image from the US Geological Survey 2011.

Fig. 4 .
Fig. 4. Monthly variations of surface and 2-meter depth fluorescence concentrations in the water at all sampling stations (mean of all sampling stations).RFU: relative fluorescence units.

Fig. 5 .
Fig. 5. Monthly variations of Pyrodinium bahamense and Ceratium furca densities in the water at all sampling stations (mean of all sampling stations and depths).

TABLE 1
Minimum, maximum and mean values of surface and bottom samples; and minimum, maximum and grand mean of the resulting mean values of surface and bottom samples, of biological and physico-chemical parameters Min=minimum, Max=maximum, SD=standard deviation, n=number of sampling periods, orgs.=organisms,RFU=relative fluorescence units.Secchi depth was measured vertically in the water column, therefore surface and bottom values are not applicable.Fig. 2. Monthly variations of surface and 2-meter depth salinity concentrations in the water at all sampling stations (mean of all sampling stations).psu: practical salinity units.