Small-Scale Biophysical Interactions and Dinophysis Blooms: Case Study in a Strongly Stratified Chilean Fjord

Abstract

Diarrhetic shellfish poisoning (DSP) toxins and pectenotoxins (PTXs) produced by endemic species of Dinophysis, mainly D. acuta and D. acuminata, threaten public health and negatively impact the shellfish industry worldwide. Despite their socioeconomic impact, research on the environmental drivers of DSP outbreaks in the Chilean fjords is scanty. From 22 to 24 March 2017, high spatial–temporal resolution measurements taken in Puyuhuapi Fjord (Northern Patagonia) illustrated the short-term (hours, days) response of the main phytoplankton functional groups (diatoms and dinoflagellates including toxic Dinophysis species) to changes in water column structure. Results presented here highlight the almost instantaneous coupling between time–depth variation in density gradients, vertical shifts of the subsurface chlorophyll maximum, and its evolution to a buoyancy-driven thin layer (TL) of diatoms just below the pycnocline the first day. A second shallower TL of dinoflagellates, including Dinophysis acuta, was formed on the second day in a low-turbulence lens in the upper part of the pycnocline, co-occurring with the TL of diatoms. Estimates of in situ division rates of Dinophysis showed a moderate growth maximum, which did not coincide with the cell density max. This suggests that increased cell numbers resulted from cell entrainment of off-fjord populations combined with in situ growth. Toxin profiles of the net tow analyses mirrored the dominance of D. acuminata/D. acuta at the beginning/end of the sampling period. This paper provides information about biophysical interactions of phytoplankton, with a focus on Dinophysis species in a strongly stratified Patagonian fjord. Understanding these interactions is crucial to improv predictive models and early warning systems for toxic HABs in stratified systems.

1. Introduction

Climate change and harmful algal blooms (HABs) of toxin-producing microalgae (toxic HABs) have drawn considerable attention in the last two decades due to the severity of their socioeconomic impacts on the shellfish industry and caged fish aquaculture [1,2]. Toxic HABs in Chile have followed the global trend of “apparent” increase [3], becoming a main threat to public health and aquaculture exploitations in the Patagonian fjords system [4]. Diarrheic shellfish poisoning (DSP) events caused by endemic species of the genus Dinophysis, mainly D. acuta and D. acuminata complex, producers of lipophilic toxins—DSP toxins (okadaic acid (OA) and dinophysistoxins (DTXs)) and pectenotoxins (PTXS)—jeopardize the flourishing shellfish industry in the Patagonian fjords system [5,6,7]. This region, which includes Los Lagos Inland Sea (X Region), is the site of more than 99% of a national mussel production of 40 × 10⁴ t per year [8].

Low biomass blooms (10²–10³ cells L⁻¹) of several species of Dinophysis may cause DSP outbreaks. Differing toxin profiles in particular strains of the same species may have very different impacts on shellfish intoxication events [9]. Dinophysis acuta and species of the D. acuminata complex exhibit marked differences in their spatio-temporal distribution in southern Chile [5,6,10,11,12] and in European Atlantic coastal waters, particularly in the western Iberian Peninsula [13,14,15,16], western Scotland [17] and southwestern Ireland [18,19].

Despite evidence of high recurrence of Dinophysis events in the Chilean Patagonia fjords system during the last four decades [5,6,10,20,21], interactions between biological processes in Dinophysis populations—growth, vertical migration, or toxin production—and quick changes of water column structure in this strongly stratified system have been poorly explored [11,12,22]. This information is crucial for the development of realistic biophysical models [23,24,25,26,27].

The specific growth rate, µ, is one of the most important intrinsic parameters in the growth equation. Nevertheless, it is extremely difficult to obtain in situ estimates of the intrinsic growth of a single species belonging to a multi-specific community. Methods acceptable for diatom species, e.g., in situ and on-board incubations or microcosm experiments, may probe detrimental and even inhibit growth in some flagellate species extremely sensitive to physical disturbance [28]. Modelers often use approximate growth-rate values derived from single and multifactorial in vitro experiments. These experimental designs are far from reproducing the field scenario, in particular, the effects of short-term hydrodynamic processes modulated by tides and wind and their interaction with microalgal swimming behaviour, daily vertical migration (or the lack of it), and active aggregation around vertical gradients [29].

In population dynamics of Dinophysis species, estimates of in situ division rates, derived from the model of Carpenter and Chang [30] and based on the mitotic index approach [31], have been successfully applied in different coastal environments. These include the Galician Rías Bajas in the northern limit of the Canary Current Upwelling System [29,32,33,34], the western Mediterranean Sea [28,35], and the southwestern coast of Ireland by the Celtic Sea [36]. Vertical heterogeneities of this crucial parameter have been observed during stratified conditions when bottle or pump samples collected at depth replaced the integrated water column sampling with vertical net tows [29,32,36].

Fjords in the Chilean Patagonia are among the most stratified systems in the world [37]. Abrupt salinity gradients all year round within the top 5–10 m is a common feature in these fjords. Nevertheless, high vertical resolution measurements of physical and biological parameters in the region have been poorly explored [11].

Dinophysis species are rare protists with a patchy distribution that affects their detection and the accuracy of conventional quantitative methods used in monitoring centres [38]. These effects may be exacerbated in highly heterogeneous systems, as is the case in the Patagonian fjords. There, multiple micro-habits promote development and niche segregation of different species [6,10,20,39]. Thus, understanding the mechanisms involved in this short-term variability is crucial for the development of early warning systems and improved risk assessment of shellfish poisoning and other hazardous events in southern Chile. This paper illustrates the small-scale variability of physical and chemical properties and the coupled response of phytoplankton distribution during an early autumn mini-cruise in Puyuhuapi Fjord, Patagonia. Special attention is paid to the microstructure and vertical gradients promoting aggregation, thin-layer formation, and growth of two toxic species of Dinophysis differing in environmental preferences.

2. Materials and Methods

2.1. Study Area

Puyuhuapi Fjord (Figure 1), Chilean Patagonia, is one of the most extended fjord and channel regions in the world [37]. This system is characterized by its abrupt bathymetry, complex coastal morphology, and marked water column stratification determined by freshwater inputs and seasonal and latitudinal patterns in precipitation [40,41]. Precipitation includes heavy rainfall (average 2700 mm y⁻¹, max. 5000 mm in exceptional years) and melt-water [40,42]. Rainfall is largely produced by the synoptic system cyclones [43] and atmospheric rivers [39] embedded in the westerly wind belt affecting southern Chile year-round [44,45].

Puyuhuapi Fjord (100 km) has two connections with oceanic waters through Moraleda Channel in the mouth and Jacaf Channel close to the head [41]. Freshwater inputs come from riverine inflows and rainfall. The main river flowing into Puyuhuapi is the Cisnes River (average discharge 218 m³ s⁻¹), located in the middle part of the fjord [41]. The estuarine (EW) layer has been subdivided, on the basis of salinity, into three ranges labeled as fresh water (FW, salinity < 11), estuarine fresh water (EFW, salinity: 11–21), and estuarine saline water (ESW, salinity: 21–31) [46]. All the above characteristics affect the fjord hydrodynamic conditions, including stratification and water residence time, which may reach values ~250 days [47,48] and promote microalgal bloom formation [12,39].

2.2. IFOP Monitoring Program

Monthly reports of phytoplankton distribution and CTD (conductivity, temperature, and depth) profiles in Puyuhuapi Fjord at a fixed sampling station (Figure 1) from September 2016 to August 2017 were obtained from the Chilean Monitoring Program at IFOP (Instituto de Fomento Pesquero). This monitoring sampling station is located close to the oceanographic buoy platform (10 km), where biophysical experiments were carried out.

Water samples for quantitative analyses of phytoplankton were collected with a dividable hose sampler (0–10, 10–20 m) [49] similar to that recommended in 1986 (also known as Lindahl’s hose sampler) by the “ICES-Working Group on Phytoplankton and Management of their Effects” to sample patchy populations of flagellated microalgae. These swimming microalgae may easily escape detection with conventional sampling with oceanographic bottles cast every 3 to 5 m [38]. Samples were immediately fixed on board with acidic Lugol’s iodine solution (Lugol samples hereinafter) to a final concentration of 0.5% [50]. For quantitative analyses, 10 mL of these unconcentrated Lugol samples were left to sediment overnight and analyzed under an inverted microscope (Olympus CKX41) using the Utermöhl method [51], as detailed in Section 2.4.1. To enumerate large species, such as Dinophysis spp., the whole surface of the chamber was scanned at a magnification of ×100 so that the detection limit was 100 cells L⁻¹.

2.3. Field Cruise Sampling

An intensive 48 h sampling was carried out during early austral autumn from 22 to 24 March 2017 in Puyuhuapi Fjord, northern Patagonia, at a 190 m deep fixed station close to a buoy from the COPAS Sur-Austral program (44°35′17.1″ S–72°43′37.5″ W) in a northern section of the fjord (Figure 1).

2.3.1. Hydrography and Turbulence Measurements

Vertical profiles of temperature, salinity, and fluorescence were obtained with an AML Oceanographic CTD profiler, model Metrec-XL (http://www.amloceanographic.com; Victoria, BC, Canada). The CTD was also equipped with an optical sensor for dissolved oxygen (DO) and other sensors for turbidity, fluorescence, and pH, which were able to measure down to 500 m deep with a sampling rate of 2 (24 measurements by seconds). The fluorescence profiles (calibrated to Chl-a equiv.) were used as a proxy for the presence of thin layers of phytoplankton (TLP) according to the criteria of Dekshenieks, et al. [52].

Measurements of turbulence microstructure were made using the Self Contained Autonomous MicroProfiler (SCAMP) manufactured by Precision Measurement Engineering Inc. (www.pme.com; San Diego, CA, USA). The SCAMP profiler recorded data at a speed of 100 Hz (100 measurements per second) with a descending free fall speed of ~10 cm s⁻¹. This instrument was equipped with a fast conductivity sensor (accuracy of ±5% of the full conductivity scale) and a fast temperature response sensor (accuracy of 0.01 °C). The measurements of vertical gradients of temperature were used to calculate the turbulent kinetic energy dissipation rate (ε) by applying the Batchelor spectrum [53,54]. Profiles of physical variables and turbulence were obtained every 2–3 h.

2.3.2. Phytoplankton and Physiological Conditions

During the cruise, unconcentrated seawater samples (125 mL) for quantitative analyses of microphytoplankton (every 4 h during the 48 h study) were collected using a 5 L Niskin bottle (www.generaloceanics.com; Miami, FL, USA) at six fixed depths (0, 5, 10, 15, 25, and 50 m; n = 75 samples). Additionally, water samples at 2 m intervals from surface to 22 m depth were collected at 09:00 h (local hour) on 23 and 24 March to study the fine-scale vertical distribution of microphytoplankton with a focus on Dinophysis spp.; 3 L were concentrated on board through 20 μm nytex filters (www.generaloceanics.com; Miami, FL, USA) and resuspended in 50 mL of seawater and fixed with formalin, to enumerate cell cycle phases of D. acuta (concentration factor = 60; n = 24 samples).

For division rate estimates, vertical net tows (0–20 m) with a 20 μm mesh net were collected and sieved through a 150 μm mesh to eliminate large micro-zooplanktonic organisms. Samples were taken every 2 h from 12:00 h to 04:00 h the next day and every hour from 04:00 h to 12:00 h (n = 33 samples), which is the time window before dawn when phased division of Dinophysis acuta is observed, and proportion of dividing and recently divided cells change very rapidly [34].

Vertical tows (0–20 m) with a net 20 µm mesh size were also collected for lipophilic toxin analyses following the same frequency described for division rate estimates (n = 33 samples); 50 mL of the towed material was collected in a Falcon tube and all tubes placed in the vessel’s deep-freeze and later transported in a portable fridge before final storage in the laboratory at −20 °C until analysis.

2.4. Sample Analyses

2.4.1. Microphytoplankton Analysis

For quantitative analyses of microphytoplankton, 10 mL of unconcentrated acidic Lugol’s-fixed samples were left to sediment overnight and analyzed under an inverted microscope (Olympus CKX41) using the method described in Utermöhl [51]. This time of ensures that all the cells can settle to the bottom of the chamber. To enumerate large species, such as Dinophysis spp., the whole surface of the chamber was scanned at a magnification of ×100 so that the detection limit was 100 cells L⁻¹. To estimate cell densities of the smaller and more abundant species, the whole surface of two diametrical transects was counted at ×250 magnification. Microphytoplankton specimens were identified at genus or species levels when possible using the worldwide recognized taxonomic guide of Thomas in addition to a valuable guide from Chilean experts [55,56]. The latter illustrates important morphological features of the Patagonian strains of cosmopolitan species or from species of similar latitudes.

2.4.2. Division Rates Estimation of Dinophysis Cells

In situ division rates were estimated, with a “post-mitotic index approach”, from the frequency of dividing (paired) and recently divided (incomplete development of the left sulcal list) cells, which were recognized by their distinct morphology as described in [34], following the model of Carpenter and Chang [30]:

$$\mu ={\displaystyle \frac{1}{n(T_c+T_r)}}\sum _{i=1}^n{{(t}_s)}_i\ln \left[1+f_c\left(t_i\right)+f_r\left(t_i\right)\right]$$

(1)

where µ is the daily average specific division rate, f_c(t_i) is the frequency of cells in the cytokinetic (or paired cells) phase (c), and f_r(t_i) is the half frequency of cells in the recently divided (incomplete development of the left sulcal list) (r) phase in the ith sample. T_c and T_r are the duration of the c and r phases, considered “terminal events” (sensu [30]) in this work; n is the number of samples taken in a 24 h cycle, and t_s is the sampling interval in hours. The duration of the selected terminal events, T_c + T_r, was estimated as the interval of time necessary for a cohort of cells to pass from one phase to the next; in this case, the time interval between the time t₀—when the frequency of cells undergoing cytokinesis, ƒ_c, is maximum—and the time t₁ when the fraction of recently divided cells ƒ_r is maximum:

$${\displaystyle \frac{1}{2}}\left(T_c+T_r\right)=(t_0-t_1)$$

(2)

where T_c, T_r, t₁, and t₀ are calculated after fitting a 5th-degree Gaussian function to the frequency data.

Vertical distribution of the lower bound of division rates (µ_min) at peak-division time (08:00 GMT) was estimated from two vertical profiles on 23 and 24 March (09:00 h), respectively. The ‘maximum frequency approach’ [31] was used to estimate µ_min at each depth z, (µ_min)_z:

$${\mu }_{min}=\ln (1+f_{max})$$

(3)

where f_max is the frequency of dividing cells at each depth. This approach assumes that all cells that divide in a given day can be recognized as undergoing or recently completed mitosis in one single sample. This approach requires collecting the sample at the time window when maximal division is expected. To estimate this parameter, between 200 and 300 cells of D. acuta were examined from each sample (vertical tows and concentrated bottle samples).

2.4.3. Toxins Sample Extraction

The phytoplankton samples were concentrated through centrifugation (20,000× g; 20 min) to obtain a final volume of 1 mL. To extract lipophilic toxins, each sample was mixed with 1 mL of methanol (100%) and then sonicated with an ultrasonic cell disruptor Branson Sonic Power 250 (Danbury, CT, USA). The extract obtained was clarified by centrifugation (20,000× g; 15 min) and then filtered through 0.20 μm Clarinert nylon syringe filters (13 mm diameter) (Bonna-Agela technologies, Torrance, CA, USA). To analyze free okadaic acid (OA) and other lipophilic toxins, an aliquot of 0.5 mL of each sample was placed in an amber vial and stored at −20 °C until analysis. For the detection of esterified OA-group toxins, an aliquot of 0.5 mL of each sample was subjected to alkaline hydrolysis following the standard procedure of the EU Reference Laboratory for Marine Biotoxins [57]. Finally, the samples were placed in an amber vial and stored at −20 °C until analysis.

2.4.4. Toxin Detection and Quantification

The presence of lipophilic toxins in the extracts was performed by LC-HRMS following the method described by Regueiro, et al. [58] but modified in order to use a shorter column and to allow enough time for the elution of all lipophilic toxins. The instrumental analysis was developed using a Dionex Ultimate 3000 UHPLC system (Thermo Fisher Scientific, Sunnyvale, CA, USA). A reversed-phase HPLC column Gemini NX-C18 (50 mm × 2 mm; 3 μm) with an Ultra Guard column C18, both from Phenomenex (Torrance, CA, USA), was used. The flow rate was set to 0.35 mL min⁻¹, and the injection volume was 10 μL. The mobile phase was used in gradient mode as follows: 85% of eluent A (100% water containing 6.7 mM NH₄OH) and 15% of eluent B (90% acetonitrile:10% water with 6.7 mM NH₄OH) was held for 1 min, followed by a linear increase to 80% B for 2.85 min and then an increase to 85% B for 0.15 min, 90% B for 0.75 min and 100% B for 3.25 min. Finally, the gradient returned to initial conditions over 2 min, and the column was re-equilibrated for 1 min. The detection of lipophilic toxins was carried out by a high-resolution mass spectrometer Q Exactive Focus equipped with an electrospray interphase HESI II (Thermo Fisher Scientific, Sunnyvale, CA, USA). The HESI was operated in negative ionization mode with a spray voltage of 3 kV and in positive ionization mode with a spray voltage of 3.5 kV. The temperature of the ion transfer tube and the HESI vaporizer was set at 200 and 350 °C, respectively. Nitrogen (>99.98%) was employed as sheath gas and auxiliary gas at pressures of 30 and 4 arbitrary units, respectively. The data were acquired in Selected Ion Monitoring (SIM) and data-dependent (ddMS²) acquisition mode. All the analyses were performed with a mass inclusion list, including the precursor ion masses, expected retention time window, and collision energy (CE) for each toxin (

Table A1. Mass spectrometry and chromatographic parameters.

Compound	Abbreviation	Chemical Formula	Molecular Ion	m/z Calculated	CE
Okadaic acid	OA	C44H68O13	[M-H]−	803.4587	55
Dinophysistoxin-1	DTX-1	C45H70O13	[M-H]−	817.4744	55
Dinophysistoxin-2	DTX-2	C44H68O13	[M-H]−	803.4587	55
Yessotoxin	YTX	C55H82O21S2	[M-2H]2−	570.2322	34
homo-Yessotoxin	homoYTX	C56H84O21S2	[M-2H]2−	577.2401	34
Azaspiracid-1	AZA-1	C47H71NO12	[M + H]+	842.5049	34
Azaspiracid-2	AZA-2	C48H73NO12	[M + H]+	856.5206	34
Azaspiracid-3	AZA-3	C46H69NO12	[M + H]+	828.4893	40
Gymnodimine	GYM	C32H45NO7	[M + H]+	508.3421	34
Pinnatoxin-G	PnTX	C42H63NO7	[M + H]+	694.4677	50
13-desMe-SPX C	SPX-1	C42H61NO7	[M + H]+	692.4521	30
Pectenotoxin-2	PTX2	C47H70O14	[M + NH4]+	876.5104	30

in Appendix A). In SIM mode, the scan mass range was set at m/z 100–1000 with a mass resolution of 70,000, the automatic gain control (AGC) was set at 5 × 10⁴, and the maximum injection time (IT) was 3000 ms. For ddMS², the mass resolution was set at 70,000, AGC at 5 × 10⁴, and IT at 3000 ms. In both cases, the isolation windows were 2 m/z. The toxin concentration in the extracts was quantified by comparing the area or the peaks obtained in the chromatograms with those of certified reference materials obtained from the NCR, Canada.

2.5. Statistical Analysis

A statistical analysis, based on the identification and counting of 24 water samples (fine-scale sampling carried out at 09:00 h on 23 and 24 March), was employed to evaluate significant differences in the phytoplankton community structure between the top stratified layer (0–10), which included the maximal density gradients of the pycnocline region and the lower mixed layer (10–20 m) below the pycnocline. Non-metric multidimensional scaling (NMDS) based on the Bray–Curtis distance matrix was used to assess the similarity of the phytoplankton community structure within the two water masses [59]. To normalize the distribution, data were first log-transformed to [LN(x + 1)] to eliminate zero values. NMDS is appropriate for the analysis of ecological community data, which may contain non-normal or discontinuous scales [60]. Analysis of similarities (ANOSIM) was applied to determine differences between the two water masses; coefficients of dissimilarity and phytoplankton species responsible for the clustering were determined with the similarity percentages (SIMPER) method. All statistical analyses and graphic representations were performed using Ocean Data View [61] and the statistical and programming software R 2.1.12 [62], packages “vegan” and “ggplot2”, available through the CRAN repository (www.r-project.org, accessed on 10 March 2024).

3. Results

3.1. Seasonal Changes in Hydrographic Conditions and Dinophysis Populations in 2016–2017

The vertical distribution of seawater temperature showed clear signals of the annual cycle (Figure 2A). A shallow (<5 m) and sharp thermocline (range 14 °C to 17.2 °C) developed near the surface layer, but the 14 °C isotherm deepened to 15 m in late summer (end February). In late autumn (May), SST declined to 8 °C, cooling of the water column reached the maximal depth in August, and thermal inversion was observed between May and August. Salinity records showed the inflow of estuarine water, including the presence of FW (S < 11) in the top 5 m in spring–summer (Figure 2B). Between May and September 2017, the estuarine salty water (ESW), limited by the isohaline of 31, was observed down to 50 m. Below the ESW, the Modified Subantarctic Water (SAAW), with salinity 31–33, formed an intermediate layer between the estuarine water and the subjacent oceanic water. The distribution of chlorophyll-a (Chl-a) in the top 20 m, with a maximum of around 8 μg L⁻¹, also showed the annual cycle signal, with lower/higher concentrations during austral fall–winter/spring–summer (Figure 2C). Vertical profiles of dissolved oxygen, with concentrations ranging from 4 to 8 mL L⁻¹ (Figure 2D), indicated good water column ventilation.

Dinophysis acuminata and D. acuta were the most abundant species of their genus. Integrated water column (hose sampler) counts showed that the two species had a bimodal distribution but distinct growth seasons. Dinophysis acuminata was detected from spring to early winter, i.e., practically all the time. Maximal cell densities (1.0 × 10³ cells L⁻¹) were found in early summer (January) and early autumn (April 2017) (Figure 3). D. acuta, detected between summer (December) 2016 and April 2017, had a shorter season and its cell maxima (3.7 × 10³ cells L⁻¹) during the last week of April (Figure 3). Ten days after the cruise, the observed densities of D. acuminata and D. acuta at the same fixed monitoring station were 5.0 × 10² cells L⁻¹ and 1.8 × 10³ cells L⁻¹, respectively.

3.2. Hydrography, Chlorophyll-a, and Nutrients during the 48 h Study

The hydrographic measurements revealed a two-layered water column structure from the surface to 50 m (Figure 4). A warmer (14.6–16.9 °C) and fresher (salinity range, 15–21) water layer occupied the top 5 m. Below, a colder (T, 10–14 °C) and saltier (S, 21–32.7) water layer extended from 5 to 50 m (Figure 4A,B). In terms of water masses, the estuarine water layer (0–15 m) was formed by the Cisnes River discharge and the subjacent intermediate layer, the Modified Subantarctic Water (MSAAW), from the mixing between the estuarine water and the oceanic subantarctic water mass (SAAW, salinity range 33–33.9). The thermohaline conditions at the surface layer contributed to generating a sharp density discontinuity at 5 m that was persistent through the whole sampling period (Figure 4C). Dissolved oxygen (DO) and pH records showed similar two-layered structures with high DO/pH at the surface and lower at the subsurface layers (Figure 4D,E).

The distribution of fluorescence (a proxy for phytoplankton biomass) indicated the formation of a subsurface chlorophyll maximum layer (SCML), which became a thin layer with values up to ~20 µg L⁻¹ congregated in a narrow band below the pycnocline (Figure 4F).

Measurements of oceanographic variables recorded from the buoy deployed at 1 m showed abrupt changes during the two tidal cycles of measurements. Near the end of the study, at dawn on March 24, Chl-a concentrations had risen (Figure 5B) to a maximum of ~20 µg L⁻¹. Increased Chl-a coincided with nitrate exhaustion due to phytoplankton consumption and a gradual decline of DO and pH values from 7.4 to 6.4 mL L⁻¹ and from 9.1 to 8.9, respectively (Figure 5C). Daily changes in salinity (between 11 and 14) and temperature (15 °C to ~17 °C) were most pronounced at the surface of brackish water (Figure 5D).

3.3. Turbulence

The turbulent kinetic energy dissipation rate (ε) presented a patchy distribution during the two days of sampling (Figure 6). Maximal values of turbulence (ε > 10⁻⁵ W kg⁻¹) were persistent between 4 and 6 m from 02:00 to 10:00 on 23 March (second day) and in the top 2 m from 18:00 to 08:00 (third day). A low-turbulence layer (ε range 10⁻⁹–10⁻⁸ W kg⁻¹) appeared between 4 and 6 m during the second part (24 h) of the study. This low-turbulence layer corresponded to the pycnocline underlying the maximal static stability region (Figure 4C).

3.4. Distribution of Dinophysis during the 48 h Cycle

Moderate cell densities of D. acuta (max. 1200 cells L⁻¹) and D. acuminata (max. 600 cells L⁻¹) were mainly restricted to the upper layer (0–10 m), although populations of the two species were vertically segregated (Figure 7).

Dinophysis acuta maximum (1.2 × 10³ cells L⁻¹) was detected at 5 m, at 12:00 h on 23 March (Figure 7A). There was no clear-cut evidence of diel vertical migration, but the maximum cell density (800 cells L⁻¹) detected at 10 m during the first half part of the cycle was closer to the surface, with the maximum (1200 cells L⁻¹) at 5 m during the second half (Figure 7A). D. acuminata was present all through the cycle, but in very low numbers. A maximum of 600 cells L⁻¹ was recorded at 10 m on March 23 at 08:00 h (Figure 7B). P. rotundatum showed a distribution pattern similar to that of D. acuta but with low cell densities throughout the cycle, with a maximum of 600 cells L⁻¹ at 10 m on March 22 at 20:00 h (Figure 7C).

Higher-resolution observations (CTD cast) of the vertical profiles of fluorescence (a proxy for phytoplankton biomass) and phytoplankton samples every 2 m from 0 to 20 m showed two interesting patterns: (i) a two-peaked vertical profile of fluorescence (one peak above (~4.5 m) and one below the pycnocline (~8 m); (ii) these peaks met the criteria to be considered a thin layer, had a thickness of ~2.8 m and their vertical position was modulated by the semidiurnal tidal regime. Phytoplankton analyses confirmed that the more intense thin layers at the base of the pycnocline were dominated by diatoms (Figure 8).

On 23 March, a well-defined thin layer, with a maximal intensity of 21 µg L⁻¹, developed at 8 m (Figure 8A). This TL coincided with a diatom maximum of 1.7 × 10⁶ cells L⁻¹ (Figure 8B). Dinoflagellates with higher cell densities (max. 1.2 × 10⁴ cells L⁻¹) were located within the pycnocline without showing a marked well-differentiated cell maximum (Figure 8C).

On 24 March, two prominent fluorescent peaks were observed. The upper and less intense peak at 4.5 m (Chl-a maximum of 18.5 µg L⁻¹) was in the upper limit of the pycnocline. The lower and more intense peak at 8 m and maximal intensity of 21.5 µg L⁻¹ was just below the lower boundary of the pycnocline (Figure 8F). The vertical distribution of diatoms did not show a thin-layer pattern, and their maxima did not exceed 7 × 10⁵ cells L⁻¹ (Figure 8G). The upper Chl-a maximum coincided with the dinoflagellates cell maxima, which included the two HAB species, D. acuminata (300 cells L⁻¹) and D. acuta (1700 cells L⁻¹) target of this study (Figure 8H–J).

An nMDS ecological community analysis supported marked differences in the distribution patterns of phytoplankton assemblages between the upper (0–10 m) stratified and the lower (10–20 m) mixed layer, with a clear separation between the two layers (stress = 0.10; Figure 9A). ANOSIM analyses also revealed significant between-site differences in cyst assemblage similarity (R = 0.98, p < 0.05; Figure 9B), which could be attributed to the relative contributions of the different species at each layer, as determined in the SIMPER analysis (Figure 9C–E). The largest percentages of dissimilarity in community structure between layers (over 40%, Figure 9C) were mainly due to the specific contribution of the two dominant diatom taxa, Chaetoceros spp. and Sundstroemia setigera.

3.5. Estimates of µ during the 48 h Study

During the intensive 48 h sampling (22–24 March), D. acuta exhibited a phased cell division, i.e., all the individuals undertaking division did it within the same time window. The frequency of dividing (fc) and recently divided (fr) cells was observed from 12:00 to 16:00 h on 22 March and from 4:00 to 12:00 h on the second and third days (23 and 24 March) (Figure 10A). Differences were observed between fc values from the second and third day. On 23 March, fc values formed a plateau from 6:00 to 12:00 h, and the peak of recently divided cells occurred 1 h after the maximum of paired cells. On 24 March, when higher division rates occurred (µ = 0.30 d⁻¹; µ_min = 0.13 d⁻¹), a sharp peak was recorded at 9:00 h, two hours after the maximum of recently divided cells (fr) (Figure 8J).

Vertical distribution of µ_min, i.e., the lower bound of the division rate estimates, at depth (µ_min)_z on 23 March and 24 March were different. On the first day, cells undergoing division were found at a depth of 2 to 25 m. High values of µ_min (>0.15 d⁻¹) were observed at the top 4 m with µ_min = 0.30 d⁻¹ at 2 m (Figure 10B). On the second day, cells undergoing division were found at a depth of 6 to 20 m. A maximum of µ_min = 0.25 d⁻¹ was recorded at 6 m (Figure 10C).

3.6. Lipophilic Toxins in Plankton

LC-MS analysis revealed the presence of okadaic acid (AO), Dinophysistoxin-1 (DTX1), and Pectenotoxin-2 (PTX2) in most plankton samples in which Dinophysis acuta, D. acuminata, and Phalachroma rotundatum were present. A first chromatogram showed a peak with a retention time of 5.77 min corresponding to the ion [M-H]⁻ 803.45 m/z. The fragmentation mass spectrum of the 803.45 ions confirmed the identification of OA since the characteristic fragment 255.12 m/z was obtained (Figure A1A,B). OA was detected in 91% of the samples and was the dominant toxin in 81% of them, with a maximum concentration of 12.28 ng mL⁻¹ in a sample obtained on 22 March at 12:00 (Figure 11). Analyses showed a second chromatographic peak with a retention time of 6.56 min and an [M-H]⁻ ion 817.46 m/z. The fragmentation spectrum showed the major production at 255.12 m/z, confirming the presence of DTX1 (Figure A1C,D). DTX1 was detected in 67% and dominated 3% of plankton samples with a maximum concentration of 5.35 ng mL⁻¹ on March 22 at 12:00 (Figure 11). Finally, the last chromatographic peak detected at 8.21 min with a parent mass [M+NH₄]⁺ 876.50 m/z corresponded to PTX2. The presence of this toxin was confirmed by the characteristic MS/MS fragment at m/z 841.46, 823.46, 805.44, and 787.43 m/z (Figure A1E,F). This toxin was present in 76% and dominant in 15% of samples, with a maximum concentration of 13.66 ng mL⁻¹ in the same sample containing the maximum concentration of OA and DTX1 (Figure 11).

4. Discussion

Coupling small-scale physical processes, such as turbulence, shear, and advection, with biological behavior (vertical migration, aggregation, physiological traits) is crucial to understand the vertical distribution of phytoplankton cells and their toxins, bloom dynamics, and patterns of toxicity [63]. Notwithstanding their importance, the study of these biophysical interactions has been constrained by the lack of adequate technologies to measure them [64]. Recent developments in high-resolution instruments for the measurement of seawater properties have attained a satisfactory level to explore microstructure and how the microalgae species and their assemblies respond to fast changes in water column structure by readjusting their vertical distribution [65].

The study of small-scale interactions of biological and physical processes in strongly stratified fjordic systems is particularly complex [66,67]. Phytoplankton and harmful algal blooms in the Chilean fjords, with Puyuhuapi Fjord as a model system, illustrate the complexity of these interactions [68]. There, strong salinity gradients and water column stability are affected by macroscale processes—El Niño Southern Oscillation (ENSO) and the Southern Annular Mode (SAM)—and by mesoscale and short-term climatic and meteorological variability [69,70]. Attention was focused here on small-scale processes and their interactions affecting phytoplankton distribution in Puyuhuapi Fjord, in particular, those driving the distribution of toxic species of Dinophysis.

Results from the 48 h study presented here highlight the key role of physical microstructure in strongly stratified systems as a promoter of vertical differentiation in phytoplankton distribution at a daily/hourly scale. Understanding how microstructures influence bloom initiation, maintenance, and dispersal of each particular HAB species is a key step to enhance species-specific predictive models.

4.1. Hydrography and Phytoplankton Distribution in the Study Year

Seasonal variability during the spring 2016–winter 2017 period showed relatively “normal” patterns of distribution of thermohaline properties and the two more abundant species of Dinophysis in Puyuhuapi Fjord. In other words, there were no extreme weather conditions (anomalies) preventing blooms of D. acuminata or D. acuta, and there was a moderate occurrence of the two species, according to the official HAB monitoring program from the IFOP. Nevertheless, the year maximum of D. acuta in mid-autumn (late April) was unusually late for this species in that region. This delay could be explained by the presence of FW (Freshwater, S > 11) at the surface, a condition associated with a low risk of bloom success for D. acuta [22] in the profiles obtained in January. On the contrary, the low salinity but temperature above 14 °C created a favorable scenario for D. acuminata. Thus, the window of opportunity for D. acuminata bloom initiation was in November, and for D. acuta, a brief favorable period was in December, but mainly in February–March.

It is important to bear in mind that one vertical profile per month is insufficient to draw conclusions, even more from a sampling point under the influence of the Cisnes River outflow. Furthermore, differences in sampling time related to the semidiurnal tidal cycle may provide significantly different results (see Section 4.2). Nevertheless, the high number of stations sampled by the monitoring agency provides snapshots of the mesoscale distribution of oceanographic parameters and a sketch of the seasonal distribution of HAB species in a particular year. All these provide a convenient background to determine in which phase (e.g., exponential, declining) of their seasonal growth were the target species populations at the time of the study.

Strains of D. acuminata and D. acuta from the Chilean Patagonia are adapted to grow within temperature and salinity ranges quite different from those of the same species in contrasting systems, e.g., the Galician Rías and the Irish Bays. Nevertheless, they share common patterns in their phenological differences, i.e., D. acuminata has a longer growth season (spring–autumn) than D. acuta (summer–early autumn), and their tolerance to sudden drops in temperature and salinity and increased turbulence (higher tolerance in D. acuminata) [13,15,71,72].

The unfrequent presence of FW at the surface during summer of 2017 in Puyuhuapi Fjord may be a consequence of the fjord “meteorological paradox”: the hotter the weather in spring, the colder the surface water from increased melt [73].

4.2. Vertical Gradients, Segregation of Diatoms and Dinoflagellates Cell Maxima, and Thin-Layer Formation during the 48 h Study

Hourly scale changes in water column structure associated with the complex hydrodynamics at the fixed sampling station were reflected in high time–depth variability in the distribution of physical and chemical parameters. The response of the main phytoplankton functional groups (diatoms and dinoflagellates) was coupled to this small-scale variability. The changes included a strengthening of density gradients and the formation of a low-turbulence layer in the 0–10 m layer, which included the pycnocline region, vertical displacements of the chlorophyll maximum, and clear-cut segregation of diatoms and dinoflagellates, which occupied different locations in relation to the pycnocline.

Pycnoclines are a common location for the development of the seasonal (spring–summer) subsurface chlorophyll maximum layer (SCML) in temperate latitudes [74]. Pycnocline gradients are particularly sharp and a permanent feature in strongly stratified fjordic systems, and SCML in these systems often evolves to become thin layers of phytoplankton (TLP) [75].

Thin layers (TLs) have been defined as structures that exhibit physical, chemical, and/or biological signatures that are different from the water just above and below. Thin layers of phytoplankton (TLP) are vertically compressed aggregations of phytoplankters that comply with established criteria of (i) thickness (a few cm to a few meters (<5 m) and horizontal extension (may reach tens of kilometers; (ii) intensity, cells concentration or proxy for biomass (e.g., Chl-a fluorescence) between three and 1–2 orders of magnitude higher than the layer immediately above or below and (iii) persistence, i.e., detectable at least in two consecutive samplings [76,77]. Different mechanisms have been proposed for thin-layer formation, but a broad division may distinguish between active and passive ways of TL development. Dinoflagellates and other swimmers with the ability to move vertically through the water column may aggregate around some physical (e.g., diurnal thermocline), chemical (nutricline or some info-chemicals), or biological (sexual pheromones) signal. On the contrary, TLs of diatoms are frequently derived from a subsurface chlorophyll maximum layer (SCML), where cells are driven by buoyancy forces. The SCML is located in the pycnocline and may evolve into a TL under the effect of vertical gradients of shear stress [78,79].

Results from the present study offered good examples of all the above cases of biophysical interactions. High Chl-a values (fluorescence as a proxy) in the CTD cast profiles identified a compressed SCML at the base of the pycnocline. This SCML, according to the 09:00 h CTD profiles on 23 March, evolved into a TL of diatoms, presumably by buoyancy-driven accumulation. No patterns of dinoflagellate aggregation were observed that day, probably due to the high turbulence lens (ε > 10⁻⁵ W kg⁻¹) observed in the preceding day profile. Within the next 24 h, a stable low-turbulence layer (ε ~10⁻⁹–10⁻⁸ W kg⁻¹) was formed, and dinoflagellates formed a TLP located in the upper limit of the pycnocline. D. acuta and D. acuminata cell maxima were embedded in this TL of dinoflagellates TL.

The response of dinoflagellates, in particular the two target species of Dinophysis, to different fields of turbulence is comparable to previous observations by some of the authors in field populations in the Galician Rías [29] and in vitro experiments [71]. Medium turbulence levels (ε) ~10⁻⁵ m⁻² s⁻³ did not affect the growth of the two species in a laboratory experiment with a turbulence (oscillating-grid) generating system [29], and thin layers of diatoms were observed within turbulence levels of (ε) 10⁻⁶–10⁻⁸ m⁻² s⁻³ in field populations in the Galician Rías [29]. Turbulence levels observed in the pycnocline region in Puyuhuapi Fjord during the present study were higher than those observed during early summer D acuminata blooms in Galicia.

It is worth pointing to the interactions between Chl-a maxima, tidal cycles, and sampling scales. Parameters measured with the buoy deployed at 1 m with a sampling frequency of 10 min showed a clear signal of the semidiurnal tidal cycle and large increments in Chl-a concentrations that escaped our observations in the CTD cast profiles. Another important observation is that all the relevant biophysical interactions described above took place within the 0–10 m layer. Therefore, meter-scale variability needs to be assessed, at least in some pilot stations, so that reports are made on the basis of sampling scales similar to the scales of the processes under assessment.

4.3. In Situ Division Rates of Dinophysis acuta and Toxins

In situ division rates are one of the most important parameters used in the growth model equation. Estimates of the specific growth rate (µ, d⁻¹) with a modification of the Carpenter and Chang model [30] have been successfully applied to Dinophysis populations to assess the contribution of biological and physical processes to the population size. Dinophysis exhibits a phased cell division, i.e., all individuals undergoing mitosis divide within a narrow time window. Total synchrony, i.e., the whole population dividing in a time window, is rarely observed in the field [34,36,80].

Studies on the in situ division of Dinophysis species have shown that the onset of light (dawn) triggers the phased cell division [11,22,34]. Nevertheless, different division rate estimates, as well as variable shapes of the time–frequency curves of mitotic cells, may be observed in populations of the same species and region. Frequency curves are sharp in fit and fast-dividing populations versus smooth and wide in those in slow-growing or stationary phases. For example, previous studies from the Galician Rías Baixas (NW Spain) had estimates of µ in autumn populations of D. acuta as low as 0.03 d⁻¹ when the rapid net growth was due to wind-driven advection of shelf populations into the bays. In contrast, µ estimates of 0.57 d⁻¹ were also observed in the autumn when good conditions for growth (haline stratification after heavy rain) resumed following advective processes [34,81]. Estimates of 0.57 and 0.46 d⁻¹ were obtained from summer populations of D. acuta during temperature-driven stratification in southern Ireland [36]. Differences in µ estimates have been related to abiotic factors (physical microstructure of the water column with optimal conditions for growth) and biotic intrinsic (growth phase of the Dinophysis population, physiological status of the cells, and size of the inoculum) and extrinsic (prey availability) factors [34,81]. High net growth rates in southwestern Ireland have been related to D. acuta cell aggregation driven by coastal jet flows in frontal areas [36].

In the present study, an active growth of D. acuta was observed in Puyuhuapi Fjord in early autumn 2017. The division process on March 23 and 24 began at dawn (4:00 h), and division rates of µ = 0.20 and 0.30 d⁻¹ and µ_min of 0.08 and 0.13 d⁻¹ were respectively recorded. These results contrast with those obtained during a previous 48 h sampling in midsummer at the same fjord. In that case, oceanographic conditions favoured a very active growth and high densities of D. acuta (maximum of 7 × 10³ cells L⁻¹). On that occasion, values of µ ranged between 0.57 and 0.76 d⁻¹, the latter being the highest value ever reported in field populations of D. acuta; values of µ_min were between 0.45 and 0.50 d⁻¹ [11]. In the present study, the sampling survey was carried out at the end of summer, and the environmental conditions allowed low D. acuta cell densities (maximum of 1.2 × 10³ cells L⁻¹) and moderate division rates. In this sense, the lower values on March 23 may reflect the response to the increased turbulence the previous day since this species is more sensitive to turbulence stress than D. acuminata [71]. In addition, higher differences between the average µ and µ_min values indicate lower synchronization in the division process, related to stress or just with late phases of a senescent population (i.e., stationary phase) [81].

Previous field studies with D. acuta have described a coincidence of the maximum growth rate at depth (µ_z)_max with the depth of the cell maximum [11,32,80]. In this case, it is assumed that the high cell numbers are the result of in situ growth. In this study, maximal µ_z on March 23 and March 24 were observed at 2 and 6 m depth, respectively (Figure 10B,C), likely contributing to the increased cell densities observed during the 48 h cycle. Nevertheless, those values did not match the depths where the cell maxima were observed (Figure 8E,J). It is possible that the D. acuta maxima were the result of recently advected cells in need of acclimation, as previously shown in populations of D. acuminata in the same fjord [11] and of D. acuta in southwestern Ireland [36].

Estimates of in situ division rates from Dinophysis populations are scanty, particularly in Chile. Data available from the Chilean fjords region is presented along with their relative position in stratified water columns (

Table 1. Temperature (°C), salinity (g kg⁻¹) and vertical gradient (0–10 m; ∆), in situ estimated division rates (mitotic index approach; day⁻¹), cell maxima (cells L⁻¹) and their position respect to pycnocline for D. acuminata and D. acuta in different cell cycle experiments carried out in Puyuhuapi Fjord in summer–autumn season. LS: late summer; MS: middle summer; EA: early summer; AP: above pycnocline; WP: within pycnocline; BP: below pycnocline.

Date	Season	Temp. [°C]		Salinity [g kg−1]		D. acuminata			D. acuta			Reference
		[0–10 m]	∆ T	[0–10 m]	∆ S	µ/µmin [d−1]	Cell Max [Depth]	Vertical Position	µ/µmin [d−1]	Cell Max [Depth]	Vertical Position
27 February 2019	LS	14.7–12.6	2.1	18.2–30.3	12.1	0.49/0.28	900 (6 m)	WP	0.57/0.45	6300 (6 m)	WP	[11]
28 February 2019	LS	15.6–12.1	2.6	18.1–30.6	12.5	0.54/0.30	1200 (4 m)	WP	0.76/0.5	6700 (6 m)	WP	[11]
19 February 2020	MS	12.4–14.2	1.9	6.1–28.1	22.0	0.29/0.19	3800 (4 m)	WP	-	200 (8 m)	BP	[22]
19 March 2020	LS	13.7–12.5	1.2	19.2–31.1	11.9	0.39/0.22	1500 (4 m)	WP	-	300 (4 m)	WP	[22]
23 March 2017	EA	15.8–13.2	2.6	16.4–29.3	12.9	-	600 (12 m)	BP	0.21/0.08	800 (10 m)	BP	This work
24 March 2017	EA	16.0–13.2	2.8	15.1–29.6	14.5	-	1700 (4 m)	AP	0.3/0.13	300 (4 m)	AP	This work

). This information will help modelers and biologists understand the microhabitat preferences of D. acuminata and D. acuta and identify extreme weather conditions either promoting exceptional blooms or preventing their occurrence. The density gradient patterns in the top 10 m may prove to be a useful indicator for early warning/risk assessment practices. For example, positive anomalies in salinity gradients (wet summers)/thermal gradients (dry summers) appear to be associated with exceptional growth of only D. acuminata/D. acuta that year.

Toxins corresponding to the two species of Dinophysis were detected throughout the sampling period. Net tows analyses are adequate to determine the toxin profile of the particulate toxin fraction in the filtered material. However, their quantitative value does not reflect the full production of toxins, considering that a high proportion of the toxins produced by Dinophysis may be released into the seawater and later adsorbed into extracellular organic matrices (mucilage, detritus, etc.). Particulate (cell quota) toxins are a parameter closely related to the division rate. Most previous studies have concluded that the most relevant factor affecting Dinophysis toxin content (accumulation) in post-exponential growth phases is the unbalance between growth and toxin production rates. Maximal values of cellular toxin have been found when growth stops (due to stress or population age) and toxin production continues [81,82].

5. Conclusions

During September 2016–September 2017, bimodal distributions of D. acuminata from early spring to early winter and of D. acuta from late spring to late autumn confirmed phenological differences between D. acuminata (longer growth season) and D. acuta (shorter season) similar to those observed between the same pair of species in geographically distant regions of the world.

Biophysical interactions led to the development of a subsurface chlorophyll maximum layer (SCML), which became a thin layer (TL) of diatoms at the base of the pycnocline and a second TL of dinoflagellates 24 h later, which included a thin layer of D. acuta, embedded with a low-turbulence lens above the pycnocline. Thus, two co-occurring thin layers segregated the two main functional groups based on their behavioural differences in response to water column re-structuring.

Vertical segregation of the cell maxima (1700 cells L⁻¹; 4 m depth) and the maximal in situ division rate (0.29 d⁻¹; 2 m depth) suggests that increased cell densities during this unusually late bloom of D. acuta were due to advective processes, but with a positive contribution of in situ growth. The preference of D. acuta for lower turbulence layers confirmed the already noted (here and elsewhere) lower tolerance of D. acuta to turbulence and salinity stress (including quick changes) than that of D. acuminata.

Results from this cruise improve our understanding of the short-term variability of Dinophysis populations in highly stratified systems, a key aspect for the development of operational models, and improved risk assessment of shellfish poisoning and other hazardous events. Improving prediction is a high research priority in one of the main seafood production regions worldwide that are subject to recurring toxic microalgae outbreaks.